This paper presents a common framework for designing tune mass dampers (TMD) and active mass dampers (AMD) to control the vibrations of shear buildings. This common framework is used for tuning the parameters of TMD and AMD devices in order to achieve a particular control objective. Thus, TMD and AMD can be better compared in practical applications. As an application example, two scaled TMD and AMD prototypes are tuned and installed in a two-storey building model. The parameters of the TMD and AMD are obtained by using this common framework and the same control objective. Finally, the performance assessment of these scaled prototypes is presented, highlighting the theoretical limitations and implementation problems, which can be used as practical guidelines for future real implementations.

Reinforcement learning is the process of using trialand-error with finite rewards to encourage an agent to take the optimal action when presented with the state of its environment. Our agent is a single-inverted pendulum, attached to a cart on a one-dimensional track, with a 4-dimensional continuous state space:
[x,α, x˙,α˙] – corresponding to the x-position (right positive), angle (counterclockwise positive from 0 vertical), x-velocity, and angular velocity, as indicated in Figure 1.

Recent developments in experimental anthropomorphically-driven prostheses have shown their potential as highly dexterous prosthetic devices. However, these prostheses are both unwearable and lack haptic feedback regarding antagonistic tensions. Here, we present a wearable, anthropomorphically-driven prosthesis with a built-in haptic feedback system. Two distinct control schemes were proposed and compared in a user study with N=6 able-bodied participants performing the Box and Blocks test. The first control scheme was designed to provide a more intuitive, human like actuation and relaxation of the hand, while the simpler controller was designed to reduce fatigue from sustaining EMG signals. Participants performed significantly better with lower fatigue levels while using the controller designed to be intuitive as opposed to the simpler controller. In addition, task performance with both controllers was better than reported performance with standard myoelectric prostheses. These findings suggest that there is potential utility in wearable anthropomorphically-driven prostheses, and provide support for future studies aimed at exploring the utility of haptic feedback in anthropomorphically-driven prostheses.

Current research on quadrotor modeling mainly focuses on theoretical analysis methods and experimental methods, which have problems such as weak adaptability to the environment, high test costs, and long durations. Additionally, the PID controller, which is currently widely used in quadrotors, requires improvement in anti-interference. Therefore, the aforementioned research has considerable practical significance for the modeling and controller design of quadrotors with strong coupling and nonlinear characteristics. In the present research, an aerodynamic-parameter estimation method and an adaptive attitude control method based on the linear active disturbance rejection controller (LADRC) are designed separately. First, the motion model, dynamics model, and control allocation model of the quad-rotor are established according to the aerodynamic theory and Newton–Euler equations. Next, a more accurate attitude model of the quad-rotor is obtained by using a tool called CIFER to identify the aerodynamic parameters with large uncertainties in the frequency domain. Then, an adaptive attitude decoupling controller based on the LADRC is designed to solve the problem of the poor anti-interference ability of the quad-rotor and adjust the key control parameter b0 automatically according to the change in the moment of inertia in real time. Finally, the proposed approach is verified on a semi-physical simulation platform, and it increases the tracking speed and accuracy of the controller, as well as the anti-disturbance performance and robustness of the control system. This paper proposes an effective aerodynamic-parameter identification method using CIFER and an adaptive attitude decoupling controller with a sufficient anti-interference ability.

@article{leylaz_2021,
title = {An Optimal Model Identification Algorithm of Nonlinear Dynamical Systems With the Algebraic Method},
author = {Leylaz, G.; Ma, S.; Sun, J.-Q.},
journal = {Journal of Vibration and Acoustics},
year = {2021},
month = {04},
volume = {143},
number = {2},
institution = {University of California, Merced, USA},
abstract = {This article proposes a nonparametric system identification technique to discover the governing equation of nonlinear dynamic systems with the focus on practical aspects. The algorithm builds on Brunton’s work in 2016 and combines the sparse regression with an algebraic calculus to estimate the required derivatives of the measurements. This reduces the required derivative data for the system identification. Furthermore, we make use of the concepts of K-fold cross validation from machine learning and information criteria for model selection. This allows the system identification with less measurements than the typically required data for the sparse regression. The result is an optimal model for the underlining system of the data with a minimum number of terms. The proposed nonparametric system identification method is applicable for multiple-input–multiple-output systems. Two examples are presented to demonstrate the proposed method. The first one makes use of the simulated data of a nonlinear oscillator to show the effectiveness and accuracy of the proposed method. The second example is a nonlinear rotary flexible beam. Experimental responses of the beam are used to identify the underlining model. The Coulomb friction in the servo motor together with other nonlinear terms of the system variables are found to be important components of the model. These are, otherwise, not available in the theoretical linear model of the system. },
issn = {1048-9002},
keywords = {data-driven modeling, algebraic differential estimation, sparse regression, system identification, nonlinear dynamics, flexible manipulator, dynamics, nonlinear vibration, system identification},
language = {English},
publisher = {ASME}
}

This article proposes a nonparametric system identification technique to discover the governing equation of nonlinear dynamic systems with the focus on practical aspects. The algorithm builds on Brunton’s work in 2016 and combines the sparse regression with an algebraic calculus to estimate the required derivatives of the measurements. This reduces the required derivative data for the system identification. Furthermore, we make use of the concepts of K-fold cross validation from machine learning and information criteria for model selection. This allows the system identification with less measurements than the typically required data for the sparse regression. The result is an optimal model for the underlining system of the data with a minimum number of terms. The proposed nonparametric system identification method is applicable for multiple-input–multiple-output systems. Two examples are presented to demonstrate the proposed method. The first one makes use of the simulated data of a nonlinear oscillator to show the effectiveness and accuracy of the proposed method. The second example is a nonlinear rotary flexible beam. Experimental responses of the beam are used to identify the underlining model. The Coulomb friction in the servo motor together with other nonlinear terms of the system variables are found to be important components of the model. These are, otherwise, not available in the theoretical linear model of the system.

In recent years, chronic pain in joints is the most common type of musculoskeletal pain among office workers. As the work culture in the office has been changed, it is the common demand that everyone has to work with a computer. In office, the worker has to sit continuously on a personal computer at least for 3–4 h. Most of the time, this leads to chronic pain in joints, generally after the period of few years of repeated exposure to such kind of regular engagement of user-computer interfacing, which is consequently converted into a serious health concern in the considerably large-sized office community. In this paper, an attempt has been made to provide a tool to assess the level of pain in the incumbent worker to have a reliable measure and at the disposal of medical practitioners to know the degree of pain to diagnose the severity of the health concern attached to it and have an optimum analysis which is not available even today. Around 250 samples have been taken and these cases have been categorized based on level of severity of pain, i.e. normal, medium, and severe. The results obtained show a considerable difference between mean and standard deviation values of pain level. Based on the data analysis of rectified EMG data, it has been observed that there is a notable difference in the Power Spectral Density of clean EMG signal w.r.t mean and Standard Deviation. The accuracy of the experimentally determined pain intensity level is a more reliable source for medical practitioners to make their diagnosis more objective one and this can be used as one of the measures to calibrate the pain-intensity.

Sample efficiency is one of the key factors when applying policy search to real-world problems.
In recent years, Bayesian Optimization (BO) has become prominent in the field of robotics due
to its sample efficiency and little prior knowledge needed. However, one drawback of BO is its
poor performance on high-dimensional search spaces as it focuses on global search. In the policy
search setting, local optimization is typically sufficient as initial policies are often available, e.g.,
via meta-learning, kinesthetic demonstrations or sim-to-real approaches. In this paper, we propose
to constrain the policy search space to a sublevel-set of the Bayesian surrogate model’s predictive
uncertainty. This simple yet effective way of constraining the policy update enables BO to scale to
high-dimensional spaces (>100) as well as reduces the risk of damaging the system. We demonstrate
the effectiveness of our approach on a wide range of problems, including a motor skills task, adapting
deep RL agents to new reward signals and a sim-to-real task for an inverted pendulum system.
Keywords: Local Bayesian Optimization, Policy Search, Robot Learning

In this paper, a new approach for Neuro-Fuzzy Controller (NFC) has been presented and compared to previously defined NFCs given in open literature. The proposed controller is based on an on-line Adaptive Neuro-Fuzzy Inference System (ANFIS) and meticulous analysis through simulations is performed to show its robustness. The performance of Neuro-Fuzzy Controllers (NFC) depends on controller inputs. To show the difference and superiority of the proposed controller, many studies in the open literature are examined and compared. Therefore, the advantages and disadvantages of the Neuro-Fuzzy controller are outlined and an optimum Neuro-Fuzzy controller is structured and presented. To test our developed controller for a nonlinear problem, having coupling effects, a 2 DOF helicopter model is chosen. Also to show the robustness, the controller performance which is applied to a 2 DOF helicopter is investigated and compared with other Neuro-Fuzzy controller structures. To better show NFC performance, NFC control results were compared with LQR+I. It is observed that besides being on-line adaptive for all systems, the controller developed has many priorities such as noiseless, strong stability, and better response time.

We formulate the control problem of active suspension systems (ASSs) that are underactuated as constraint-following. The objective is to drive the system to follow a series of servo constraints or control goals which include both equality constraint of the reference trajectory and inequality constraint of the boundary restriction on the sprung mass displacement. The equality constraints may be holonomic or nonholonomic. The control design is implemented in two steps. First, a constraint-following control (CFC) is applied to drive ASSs to follow equality constraints, under the premise of ignoring inequality constraints. No auxiliary variables or pseudo variables are required for the control design. Second, diffeomorphism is employed to integrate inequality constraints into equality constraints, by which a novel CFC is proposed to deal with both equality and inequality constraints. The effectiveness of the proposed approach is demonstrated by rigorous proof. Furthermore, we investigate an ASS control problem in which the equality constraint is determined by the skyhook model, and the inequality contraint of the sprung mass displacement is determined by the suspension deflection and the tire dynamic displacement. Experimental and numerical simulation results on this case are presented for demonstrations.

When learning policies for robot control, the required real-world data is typically prohibitively expensive to acquire, so learning in simulation is a popular strategy. Unfortunately, such polices are often not transferable to the real world due to a mismatch between the simulation and reality, called ‘reality gap’. Domain randomization methods tackle this problem by randomizing the physics simulator (source domain) during training according to a distribution over domain parameters in order to obtain more robust policies that are able to overcome the reality gap. Most domain randomization approaches sample the domain parameters from a fixed distribution. This solution is suboptimal in the context of sim-to-real transferability, since it yields policies that have been trained without explicitly optimizing for the reward on the real system (target domain). Additionally, a fixed distribution assumes there is prior knowledge about the uncertainty over the domain parameters. In this letter, we propose Bayesian Domain Randomization (BayRn), a black-box sim-to-real algorithm that solves tasks efficiently by adapting the domain parameter distribution during learning given sparse data from the real-world target domain. Bayesian Domain Randomization (BayRn) uses Bayesian optimization to search the space of source domain distribution parameters such that this leads to a policy which maximizes the real-word objective, allowing for adaptive distributions during policy optimization. We experimentally validate the proposed approach in sim-to-sim as well as in sim-to-real experiments, comparing against three baseline methods on two robotic tasks. Our results show that BayRn is able to perform sim-to-real transfer, while significantly reducing the required prior knowledge.

In the wake of recent trends toward the idea of smart grids, interest is growing in decentralized energy supply. An urban environment with expected low wind speeds is particularly well suited for small scale wind turbines; various experiments using sustainable micro-energy sources have proven to contribute to a leveling of fluctuation in the electric network. This work is concerned with optimizing the operation of a cross-flow vertical axis wind turbine (VAWT) using magnetic braking. A first design step involved the development of a test bench prototype with emphasis on validation of available eddy current brake models. Another design step was focused on deriving a wind turbine model, in this case also a grey box model, for simulation purposes. The model developed has been based on input-output data retrieved from a VAWT prototype in a wind tunnel. A last development step within this work involved calculation of a rough maximum braking torque for a requirement analysis, combination of the models developed previously and a concept for a two-mode control strategy governed by supervisory control.

Artificial intelligence (AI) has an increasing influence in the manufacturing and processing industries. An increasingly interesting area of application is the design of controllers for technical systems through reinforcement learning. This allows the design of a controller with reduced effort for human experts, which is normally involved in the design process, for example by eliminating the determination of suitable controller parameters. In this paper a neural controller for a rotational inverted pendulum is developed by using artificial neural networks in combination with reinforcement learning. The rotational inverted pendulum is a classic example of a nonlinear and unstable system and has been little covered in previous publications on this subject. Furthermore, the behaviour of the neural controller is compared with a conventional controller when swinging up and balancing a rotational inverted pendulum.

In this study, a new perspective for developing laser scanner rangefinder based data clustering system for a 2DOF robotic ball balancer was proposed. The study focused on detecting an object (i.e., ball) on the tilt-table robotic platform using the sensor fusion and data clustering systems proposed. Clustering system was modeled by following the principles of hierarchical clustering method. The developed system involving the clustering and sensor fusion algorithms was embedded in Matlab-Simulink environment to be able to run in real-time applications. The system was tested using an experimental platform including a 2DOF robotic ball balancer equipped with high resolution encoders and a laser scanner rangefinder. In the experiments, the goal was to detect the ball and its position not only on the flat but also on the tilted platform. A camera was also attached to the top of the experimental setup and used to monitor the location of the ball on the platform. By this way the results obtained using the proposed system could be verified for accuracy, performance and repeatability issues.

This paper is dedicated to the experimental analysis of discrete-time differentiators implemented in closed-loop control systems. To this end, two laboratory setups, namely an electro-pneumatic system and a rotary inverted pendulum have been used to implement 25 different differentiators. Since the
selected laboratory setups behave differently in the case of dynamic response and noise characteristics, it is expected that the results remain valid for a wide range of control applications. The validity of several theoretical results, which have been already reported in the literature using mathematical analysis and
numerical simulations, has been investigated, and several comments are provided to allow one to select an appropriate differentiation scheme in practical closed-loop control systems.

Motion intention detection is fundamental in the implementation of human-machine interfaces applied to assistive robots. In this paper, multiple machine learning techniques have been explored for creating upper limb motion prediction models, which generally depend on three factors: the signals collected from the user (such as kinematic or physiological), the extracted features and the selected algorithm. We explore the use of different features extracted from various signals when used to train multiple algorithms for the prediction of elbow flexion angle trajectories. The accuracy of the prediction was evaluated based on the mean velocity and peak amplitude of the trajectory, which are sufficient to fully define it. Results show that prediction accuracy when using solely physiological signals is low, however, when kinematic signals are included, it is largely improved. This suggests kinematic signals provide a reliable source of information for predicting elbow trajectories. Different models were trained using 10 algorithms. Regularization algorithms performed well in all conditions, whereas neural networks performed better when the most important features are selected. The extensive analysis provided in this study can be consulted to aid in the development of accurate upper limb motion intention detection models.

The need to design a robust, accurate and lightweight real-time control architecture for a low-cost embedded system has become a necessary step in robotics applications. Therefore, the goal of this work is to design an integrated and accurate control system for a differential wheeled mobile robot. A hierarchical control approach has been adopted in this work which, takes into account the sub-systems controller of the problem related to robot path tracking. To this end, the control structure includes the design of a sliding mode control (SMC) scheme to stabilize the robot and precisely track the path that matches the kinematics of the mechanical structure. Moreover, the control structure provides a new design of the DC motor controller for optimum performance. To obtain an integrated control system for automated tasks, an accurate position estimation and an optimized high-level path planning algorithm have been proposed. The robustness and performance of the approved control and estimation systems have been verified through real-time implementations in different experiments. The results show the high accuracy of the robot’s tracking on different paths.

It is commonly known that the design of control is difficult for the systems with non-minimum phase behavior, the coupling between the controlled variables and integral instability. The paper presents such a typical class of nonlinear system and propose a new technique generalized extended state observer-based sliding mode decoupling control. The proposed control algorithm is also able to compensate for the matched and mismatched uncertainties encountered in such nonlinear non-minimum phase systems. The decoupling control approach is used to decouple the entire system framework into subsystems. The sliding surface of the individual subsystem was composed using the estimated states of GESO. The proposed technique can nullify the effect of external disturbances, parametric uncertainties and unmodelled dynamics due to the inclusion of disturbance compensation gain in the control law. The proposed methodology enables the system to accomplish asymptotic stability and better dynamic response with lesser IAE and ITAE values than existing decoupling control strategies. An experimental application of the proposed technique to Quanser’s rotary pendulum module is investigated.

A Fuzzy Resource Manager (RM) to compensate communication loads in real-time systems is presented. The design is based on a new model of a Constant Bandwidth Server (CBS), which is responsible for assigning time slots to tasks with the highest priority when idle time is available. Assuming that each application can be executed at different service levels, without being below a minimum limit, a fuzzy approach is introduced that allows to adjust the time resources assigned to each task and to compensate non-linearities in time resources requests. The RM increases or decreases the virtual platform for each application and assigns a maximum process time budget for it, which is gradually used and refilled when depleted, without affecting the other applications.
The scheme self-adjusts to sudden changes in applications process time requirements.

@article{kana_2020,
title = {Human–Robot co-manipulation during surface tooling: A general framework based on impedance control, haptic rendering and discrete geometry},
author = {Kana, S.; Campolo, D. Tee, K.-P.},
journal = {Robotics and Computer-Integrated Manufacturing},
year = {2021},
month = {02},
volume = {67},
institution = {Nanyang Technological University, Singapore Institute for Infocomm Research, Singapore},
abstract = {Despite the advancements in machine learning and artificial intelligence, there are many tooling tasks with cognitive aspects that are rather challenging for robots to handle in full autonomy, thus still requiring a certain degree of interaction with a human operator. In this paper, we propose a theoretical framework for both planning and execution of robot-surface contact tasks whereby interaction with a human operator can be accommodated to a variable degree. The starting point is the geometry of surface, which we assume known and available in a discretized format, e.g. through scanning technologies. To allow for realtime computation, rather than interacting with thousands of vertices, the robot only interacts with a single proxy, i.e. a massless virtual object constrained to ‘live on’ the surface and subject to first order viscous dynamics. The proxy and an impedance-controlled robot are then connected through tuneable and possibly viscoelastic coupling, i.e. (virtual) springs and dampers. On the one hand, the proxy slides along discrete geodesics of the surface in response to both viscoelastic coupling with the robot and to a possible external force (a virtual force which can be used to induce autonomous behaviours). On the other hand, the robot is free to move in 3D in reaction to the same viscoelastic coupling as well as to a possible external force, which includes an actual force exerted by a human operator. The proposed approach is multi-objective in the sense that different operational (autonomous/collaborative) and interactive (for contact/non-contact tasks) modalities can be realized by simply modulating the viscoelastic coupling as well as virtual and physical external forces. We believe that our proposed framework might lead to a more intuitive interfacing to robot programming, as opposed to standard coding. To this end, we also present numerical and experimental studies demonstrating path planning as well as autonomous and collaborative interaction for contact tasks with a free-form surface. },
keywords = {Robot-surface interaction, Impedance control, Human–Robot collaboration, Virtual proxy, Virtual fixtures, Triangular mesh model, Geodesics},
language = {English},
publisher = {Elsevier B.V.}
}

Despite the advancements in machine learning and artificial intelligence, there are many tooling tasks with cognitive aspects that are rather challenging for robots to handle in full autonomy, thus still requiring a certain degree of interaction with a human operator. In this paper, we propose a theoretical framework for both planning and execution of robot-surface contact tasks whereby interaction with a human operator can be accommodated to a variable degree.

The starting point is the geometry of surface, which we assume known and available in a discretized format, e.g. through scanning technologies. To allow for realtime computation, rather than interacting with thousands of vertices, the robot only interacts with a single proxy, i.e. a massless virtual object constrained to ‘live on’ the surface and subject to first order viscous dynamics. The proxy and an impedance-controlled robot are then connected through tuneable and possibly viscoelastic coupling, i.e. (virtual) springs and dampers. On the one hand, the proxy slides along discrete geodesics of the surface in response to both viscoelastic coupling with the robot and to a possible external force (a virtual force which can be used to induce autonomous behaviours). On the other hand, the robot is free to move in 3D in reaction to the same viscoelastic coupling as well as to a possible external force, which includes an actual force exerted by a human operator. The proposed approach is multi-objective in the sense that different operational (autonomous/collaborative) and interactive (for contact/non-contact tasks) modalities can be realized by simply modulating the viscoelastic coupling as well as virtual and physical external forces. We believe that our proposed framework might lead to a more intuitive interfacing to robot programming, as opposed to standard coding. To this end, we also present numerical and experimental studies demonstrating path planning as well as autonomous and collaborative interaction for contact tasks with a free-form surface.

The steady-state operation and controlled position tracking in ball balancer applications are largely affected due to their nonlinear and underactuated behavior. To overcome these drawbacks, this paper develops a wavelet-based fuzzy controller which operates in closed loop with the system by measuring the ball position and the plate angle. The proposed approach adapts discrete wavelet transform for denoising the error signal and tuning the weights of the fuzzy controller. To test the effectiveness of the proposed approach, numerical simulations are performed by modeling a two degree of freedom ball balancer system. The eminence of the controller is assessed by comparing its operation on the modeled system with conventional fuzzy logic controller and by calculating the root mean square error, and time response analysis. The results showed a steady and precise response of the proposed approach to the framework of positioning ball on the plate.

Proximal and trust-region policy optimization methods (PPO and TRPO) belong to the standard reinforcement learning toolbox. Notably, PPO can be viewed as transforming the constrained TRPO problem into an unconstrained one, either via turning the constraint into a penalty or via objective clipping. In this chapter, an alternative problem reformulation is studied, where the information loss is bounded using a novel transformation of the KullbackLeibler (KL) divergence constraint. In contrast to PPO, the considered method does not require tuning of the regularization parameter, which is known to be hard due to its sensitivity to the reward scaling. The resulting algorithm, termed information-loss-bounded policy optimization (ILBPO), both enjoys the benefits of the first-order methods, being straightforward to implement using automatic differentiation, and maintains the advantages of the quasi-second order methods. It performs competitively in simulated OpenAI MuJoCo environments and achieves robust performance on a real robotic task of the Furuta pendulum swing-up and stabilization.

This paper presents the implementation of a position-torque control law for a 6DOF Stewart platform for tracking trajectories through its end-effector. First, the kinematics and a configuration description are depicted. Then, a control scheme is suggested and implemented using a Quanser data acquisition card that has compatibility with Matlab Simulink in order to control the linear positioners of the robot. The use of these elements merges a simple but rewarding control of this parallel robot towards a new scheme of technologic incorporation for aerospace applications which make use of the Stewart platform as a motion generator.

In this paper, meta-heuristic tuning of LQR weighting matrices using various objective functions on the experimental flexible arm under the effects of disturbance is investigated. The use of flexible and lightweight systems provides certain advantages such as high operating speeds, low electricity consumption, and low initial investment costs. However, such systems are more prone to vibration-related problems than their rigid and heavyweight counterparts. One of the closed-loop control methods used to suppress these vibrations is LQR control, but the success of the controller depends on the choice of the gain and regulator matrices. In this study, the Bees algorithm, a meta-heuristic search algorithm, is used to determine LQR matrices. The system performance is compared with similar studies in the literature. The proposed objective function reveals an improvement of 3.1% in rotor maximum overshoot compared to studies in the literature. Flexible link maximum overshoot has been reduced from 17.8 to 7.1%. In order to validate the applicability and repeatability of the proposed method under the noise and disturbance in real systems, experimental verification tests were conducted using the Quanser Flexible Link system. As a result, the proposed method has been experimentally validated and it is estimated that it may well be a very useful controller design approach in various engineering systems and related control system developments.

Development of practical control approaches for the under-actuated chaotic systems such as the robot
manipulators are challenging due to the unpredictable character of the chaotic dynamics, and the
inevitable real-time application properties like delays, saturations, and uncertainties In this paper, we
propose a model free digital adaptive control approach, which considers the time delay of the control
signal, actuator saturation, and non-parametric uncertainties, for an under-actuated manipulator. We
also develop a chaos control to learn the unbiased and smooth digital control policy inside the chaotic
regions of the continuous time under-actuated manipulator. We perform real-time experiments in
a dynamic environment with the proposed digital adaptive control. Then we compare the results of
the learning and control with and without chaos control. We observe that the proposed model free
adaptive control approach can accurately learn both the long-term predictor and unbiased control
policy even in the chaotic regions of the under-actuated robot manipulator.

The paper deals with the hierarchical modeling, simulation and model-based investigation of a nonlinear 1 DOF helicopter model, which is the most important building block of complex multi-rotor systems. The motivation behind the work is the increasing number of multi-rotor drones with various applications in diverse areas like smart cities and smart infrastructures, such as intelligent surveillance, autonomous (or semi-autonomous) flight, automatic object detection and recognition, carrying various payloads (sensors, packages) etc. In order to make these flying robots intelligent, a sophisticated control strategy is required, which is partly based on sensory information provided by the on-board sensors (gyroscope, accelerometer) - recently supported by visual data provided by on-board camera(s) - concerning the low-level feedback control loops. Most of the modern drones are built on an embedded computer, and the first layer of control is usually implemented in the firmware, thus the low level control commands are given per se. For the accomplishment of complex tasks (specific "behavior" of the drone) high level control strategies have to be implemented. In order to make these control loops efficient and robust, model based analysis of the controlled system had been carried out along with the real-life tests of the various control strategies. Since both the structure and the parameter set of the model system can be easily changed, model based study is a very efficient way of experimenting with novel control approaches.

This paper deals with the fractional-order modeling, stability analysis and control of robotic manipulators, namely a single flexible link robotic manipulator (SFLRM) and 2DOF Serial Flexible Joint Robotic Manipulator (2DSFJ). The control law is derived using Pole Placement (PP) method. This paper uses Mittag–Leffler function for the analysis of SFLRM in the time domain. The stability analysis of the fractional model is carried in a transformed Ω-Domain, and from the analysis, it is observed that the response of the fractional model of SFLRM robotic manipulator is stable. The main motive behind this analysis is to understand the fractional behavior of robotic manipulators, and it is well known from literature that most of the real-world systems have their own fractional behavior. Furthermore, a real-time SFLRM and 2DSFJ setups are considered to validate the results obtained and it is found that the control law suggested by PP method improves the settling time of the robotic manipulators.

Simultaneous large-scale recordings and optogenetic interventions hold the promise to decipher the fast-paced and multifaceted dialogue between neurons that sustains brain function. Here we developed unprecedentedly thin, cell-sized Lambertian side-emitting optical fibers and combined them with silicon probes to achieve high quality recordings and ultrafast multichannel optogenetic inhibition in freely moving animals. Our new framework paves the way for large-scale photo tagging and controlled interrogation of rapid neuronal communication in any combination of brain areas.

Compared to conventional arrested heart surgery, beating-heart surgery is promising as the advantages of eliminating adverse effects caused by a heart-lung bypass machine and enabling intraoperative evaluation of heart motion. However, the fast motion of the heart introduces a significant challenge for beating-heart surgery. In this paper, a teleoperation system, which employs an impedance control for the master robot and an ultrasound image-based position control for the slave robot (surgical robot), is proposed to achieve non-oscillatory force feedback and heart motion compensation, respectively. Specifically, an impedance model is designed for the master robot to provide the human operator (surgeon) with non-oscillatory haptic feedback. To compensate for the beating heart’s motion, ultrasound imaging is used to obtain the position of the point of interest (POI) on the heart tissue. As the use of ultrasound imaging introduces non-negligible time delay caused by image acquisition and processing, a recurrent neural network (NN)-based physiological organ motion predictor is proposed. The predicted POI position is used to control the slave robot to automatically compensate for the beating heart’s motion. The proposed method is validated through experiments. The proposed control strategy with NN-based heart motion predictor is compared to the other two strategies without heart motion predictor and with an extended Kalman filter (EKF)-based heart motion predictor. The experimental results present that the proposed strategy with NN algorithm shows significant advantages (higher synchronization accuracy and relatively steady slave-heart contact force) over the other two strategies.

Sophisticated automation is dependent upon industrial robots with precise position control servomotors. In the present work, a fractional order fuzzy PID controller (FOFPID) was designed to improve the position control response of a rotary servo system. The control errors and their fractional derivatives were applied as input scaling factors to the fuzzy logic controller (FLC). Fuzzy inference system (FIS) was employed to restrict performance indices in control signals. FOFPID performance for a Quanser servomotor was compared against PID, fractional order PID and fuzzy PID (FPID) controllers in terms of both simulations and hardware implementation. The controller performance evaluation metrics included time domain characteristics such as rise, peak, settling times and over / undershoots. The integral absolute and integral time absolute errors were also evaluated for comparisons among the different controllers. Results establish that only the FOFPID controller achieves zero percent over and undershoots. It also attains better set point tracking over other controllers. Thus, FOFPID controller is the key to precision robotics in the smart factories of tomorrow.

In this article, we formulate a framework for the networked control systems over limited channels with a novel reference input and hysteresis quantizer based triggered control (RIHQBTC) method. Both hysteresis quantization and the network induced delay are considered and the finite‐gain L2 stability is analyzed. By thoroughly exploring the structure of the hysteresis quantizer, we merge the hysteresis quantizer parameter with the reference input to construct the triggering condition. To compensate for the negative effect caused by the network induced delay, three kinds of local controllers are designed to connect the transformed controller with the original one. Moreover, the corresponding coders and decoders are proposed to bound the transmission bit rates. With this RIHQBTC method, it is proved that Zeno behavior could be intrinsically eliminated. The superiority of the proposed method is demonstrated by a numerical example and its efficiency is further verified on the wheeled mobile robot.

Fluidic soft actuators have drawn more and more research interest in recent years, but it is not clear on their behaviors of dynamical energy consumption, which is critical to power system design. To solve this problem, this work develops an analytical model of required energy including elastic potential energy and dissipated energy for fiber-reinforced fluidic soft actuators based on the theory of viscoelasticity. Since the parameters of soft material are difficult to obtain, the key parameters in the model are identified by experimental data and genetic algorithm. To verify the general performance of the developed model, a series of fiber-reinforced fluidic soft actuators with different dimensions are fabricated and tested. The average errors of the experimental value and the predicted value are equal to 0.045 J, 0.003 J, 0.001 J, and 0.011 J, which are less than 15% of the nominal values. The experimental results show that the model can capture the energy characteristics of the soft actuators under various operating conditions with expected accuracy. Finally, the energy consumption of a robotic arm composed of three soft actuators is also evaluated. This work can promote the understanding on the energy requirement of soft robots and provide further reference to the optimization design of power system.

Design and development of robotic-assistance must consider the abilities of individuals with disabilities. In this paper, a 8-DOF kinematic model of the upper limb complex is derived to evaluate the reachable workspace of the arm during interaction with a planar robot and to serve as the basis for rehabilitation strategies and assistive robotics. Through inverse differential kinematics and by taking account the physical limits of each arm joint, the model determines workspaces where the individual is able to perform tasks and those regions where robotic assistance is required. Next, a learning-from-demonstration strategy via a nonparametric potential field function is derived to teach the robot the required assistance based on demonstrations of functional tasks. The paper investigates two applications. First, in the context of rehabilitation, robotic assistance is only provided if the individual is required to move her arm in regions that are not reachable via voluntary motion. Second, in the context of assistive robotics, the demonstrated trajectory is scaled down to match the individual’s voluntary range of motion through a nonlinear workspace mapping. Assistance is provided within that workspace only. Experimental results in 5 different experimental scenarios with a person with cerebral palsy confirm the suitability of the proposed framework.

@article{zhu_2021,
title = {Robust attitude control of a 3-DOF helicopter considering actuator saturation},
author = {Zhu, X.; Li, D.},
journal = {Mechanical Systems and Signal Processing},
year = {2021},
month = {02},
volume = {149},
institution = {Southeast University, China Shanghai Maritime University, China},
abstract = {This paper presents a comparative research of robust attitude control for helicopter system. Considering the actuator saturation problem in the controller design process, two control approaches are mainly investigated. A mixed H∞ performance and linear quadratic regulator (LQR) based robust controller is proposed as the first approach, in which the control input is indirectly constrained in the cost function. While a constrained H∞ performance based robust attitude controller is developed as the other approach, in which the actuator saturation problem is directly considered in the dynamical model as well as following controller gain calculation. Based on Lyapunov stability theory, sufficient conditions in terms of linear matrix inequalities (LMI) are given. By using Quanser’s 3-DOF laboratory helicopter platform, comparative simulation and experimental tests are conducted. Payload variation and wind disturbance are all considered in the experimental test. And the results demonstrate the effectiveness as well as different performance of proposed robust attitude controllers. },
keywords = {Robust attitude control 3-DOF helicopter, Actuator saturation, Comparative experimental tests},
language = {English},
publisher = {Elsevier Ltd.}
}

This paper presents a comparative research of robust attitude control for helicopter system. Considering the actuator saturation problem in the controller design process, two control approaches are mainly investigated. A mixed H∞ performance and linear quadratic regulator (LQR) based robust controller is proposed as the first approach, in which the control input is indirectly constrained in the cost function. While a constrained H∞ performance based robust attitude controller is developed as the other approach, in which the actuator saturation problem is directly considered in the dynamical model as well as following controller gain calculation. Based on Lyapunov stability theory, sufficient conditions in terms of linear matrix inequalities (LMI) are given. By using Quanser’s 3-DOF laboratory helicopter platform, comparative simulation and experimental tests are conducted. Payload variation and wind disturbance are all considered in the experimental test. And the results demonstrate the effectiveness as well as different performance of proposed robust attitude controllers.

Fault estimation plays an important role in real-time monitoring, diagnosis and active fault tolerant control of helicopter system. This paper presents several robust observer based actuator fault estimation designs for an experimental 3-DOF helicopter system with the consideration of potential actuator saturation. In order to deal with the non-differentiable problem that induced by input saturation theoretically, the saturation function is approximated by a continuous function in the fault estimation design process. Both Luenberger observer and unknown input observer are adopted in the actuator fault estimation design scheme. Furthermore, in order to achieve sensor reduction and also avoid inaccurate measurement, simultaneous state and actuator fault estimation design is developed for the helicopter system. An energy-to-energy strategy is applied to ensure the robust performance of the proposed fault estimation approaches, in which approximation error of the saturation function, model uncertainties and external disturbances are all taking into account. In addition, pole placement technique is utilized to adjust the transient response of the estimation errors. Observer gains of the proposed fault estimation designs are derived by solving linear matrix inequalities. Finally, based on Quanser’s 3-DOF helicopter platform, comparative simulation and experimental tests are all carried out to show the effectiveness as well as performance of the proposed fault estimation approaches.

In this paper a robust tracking control strategy is proposed for Unicycle Mobile Robots (UMRs) under the influence of some disturbances. The proposed strategy is designed taking into account the perturbed kinematic model and it is based on two robust control techniques: Sliding-Mode Control (SMC) and Attractive Ellipsoid Method (AEM). The control of the heading angle is designed by means of a saturated SMC algorithm while the position control is designed by means of the AEM considering a Barrier Lyapunov function (BLF) approach. Simulation results illustrate the performance of the proposed robust controller compared to a classic UMR controller. Finally, some experimental results and comparisons illustrate the performance of the proposed strategy.

Closed-loop functional electrical stimulation (FES) control methods are developed to enable motorized assistive split-crank (i.e., a cycle without mechanical coupling between the lower limbs) cycling for rehabilitation efforts for people with lower limb movement disorders. The non-dominant side tracks a desired range of cadence and the dominant side tracks a range of position offsets centered around the position of the non-dominant side. A multi-level switched system with switched control objectives is applied to both sides of the cycle-rider system. Assistive, uncontrolled, and resistive modes for the dominant and non-dominant subsystems are based on position and cadence, respectively. Global exponential tracking to upper and lower bounds of an uncontrolled desired region is proven for each side via Lyapunov-based analysis using switched system methods. Experiments on both able-bodied participants and participants with neuromuscular conditions show the performance of the switched control system for split-crank FES-cycling. From volitional to controlled pedaling in able-bodied participants, average RMS cadence error of the non-dominant, RMS position error of the dominant,
and cadence differential between the two legs improved by 76.2%, 65.3%, and 58.0%, respectively. On average, experiments on participants with neuromuscular conditions resulted in RMS errors that were 45.8%, 92.6%, and 52.0% higher than controlled trials on able-bodied participants, but 65.3%, 33.3%, and 36.3% lower than volitional-only trials of able-bodied participants.

A novel sliding mode prediction fault-tolerant control method is designed for a class of discrete uncertain quad-rotor systems with multi-delays and actuator faults in this paper. Firstly, a quasi-integral sliding mode surface is designed as a prediction model to eliminate the sliding mode approaching mode and ensure the global robustness. Secondly, aiming at actuator faults and multiple time delays, a double-power function reference trajectory with enhanced fault compensation is designed to reduce the impact of time delays on the system, and improve fault-tolerant control accuracy. Thirdly, an inverse time coyote optimization algorithm (ICOA) is designed for rolling optimization. Based on gaining good convergence speed, the ICOA can avoid local extremes and balance local development capabilities and global search capabilities. Finally, the comparison simulation study on the quad-rotor proves the superiority of the proposed control algorithm.

For discrete systems with sensor and actuator failures, this paper proposes an observer that can estimate both sensor failures and actuator failures and designs a sliding mode predictive fault-tolerant control method based on an improved whale optimization algorithm. First, a proportional-integral observer that can observe actuator fault and sensor fault is designed to estimate the value of faults, which greatly improves the work efficiency. After that, a global sliding mode surface is designed as a prediction model, so that the initial state of the system is located on the sliding mode surface to avoid the instability of the sliding mode approaching the process. The reference trajectory of a power function with uncertainty and disturbance compensation is designed to reduce the bad influence of uncertainty and disturbance on the system and suppress chattering greatly. Meanwhile, in the rolling optimization part, an improved whale optimization algorithm(IWOA) is designed to optimize the control law. Finally, the simulation results on the four-rotor helicopter simulation platform show the practicability and superiority of the algorithm.

A microsurgery-specific haptic interface with articulated structure, possessing 3 active, 4 passive and 3 sup-plementary degrees of freedom (DOFs), was designed and developed. The system includes a 3-DOF active serial linkage design, mimicking the human upper extremity. It is combined with a microsurgery-specific end-effector, comprised of a 3-DOF passive gimbal mechanism and a 1-DOF passive exchangeable surgical toolset. The spherical gimbal workspace twinned with the larger linkage arm workspace allow accurate and delicate movements to maneuver the surgical tool in narrow corridors, corresponding to conventional surgery. The structural design of the device is based on kinematic performance measures aimed to enhance ergonomics and mitigate training limitations of novice end-users. A supplementary 3-DOF adjustable base further improved er-gonomics for end-users of different sizes. Force feedback was provided to improve safety, i.e. avoidance of force errors in execution of surgical tasks. This haptic interface provides high positional and output force resolution (0.92–1.46 mm in Cartesian coordinate system and 0.2 N, respectively), wide range of allowable force (up to 15 N), high global manipulability (0.77), global isotropy (0.68) and stiffness (2.3 N/mm). The work, indeed, in-troduces a novel kinematic performance index, Dexterous Conditioning Index (DCI), based on the volume of dexterous workspace, correlating to ease of use and decreased singularity. Accordingly, the proposed haptic device demonstrates a high level of accuracy, haptic feedback, dexterous workspace, and performance, ideal for implementation in robot-assisted microsurgery

In this paper, a super twisting observer-based full order sliding mode control is proposed for a non-linear uncertain system. The super twisting observer estimates an n−th system state and the derivative of the n−th system state. It ensures finite convergence of estimation error to zero. The proposed method retains attractive features of full order sliding mode control like finite-time convergence of system states, continuous control, and the system’s full order dynamics when the system is in the sliding mode. The scheme is validated on a 2-DOF helicopter system laboratory setup, and the performance is compared with two well-known observers, i.e., a non-linear extended state observer and a sliding mode observer-based control using two-link manipulator example.

In this paper, by deliberately introducing beneficial nonlinear stiffness and damping characteristics and employing beneficial disturbance effects, a unique switching logic-based saturated tracking control (SLSTC) scheme is proposed for active suspension control. Different from exiting robust control methods, the designed controller switches its structure to cancel out or retain the disturbance based on a novel well-designed disturbance effect indicator; and also, the inherent nonlinearity would be employed instead of directly cancelled by following the introduced bioinspired X-dynamics. The proposed SLSTC method could be for the first time to establish a deliberate assessor on the disturbance effect, and then to make full use of its positive response. These special designs are shown to be very beneficial for improving transient control response and saving control energy significantly. Experimental results validate that, the designed tracking controller can have obvious transient performance improvement up to 63% or more and save the control energy up to 51% or more, compared to traditional controllers. The results of this study would present a novel “robust and green” insight into active suspension control of vehicles by exploiting nonlinear benefits and disturbances.

Vibration and displacement control are of critical importance for both high-rise and ultra high-rise building systems. A single-floor building-like structure equipped with an active mass damper (AMD) is investigated in this paper. Optimal vibration control, while dealing with system uncertainties, is realized by the reinforcement learning (RL) technique. When the unexpected natural disasters (such as strong wind excitation) occur, the proposed controller applying to the active mass damper can compensate the increase of the system vibration caused by external disturbances. In addition, a Lyapunov candidate is used to derive a semi-global uniformly ultimately bounded (SGUUB) property. The experimental platform is mainly composed of one flexible floor and a linear cart system. Both the acceleration and the displacement responses of the floor are provided and compared separately for passive mode, proportional-velocity (PV) control and RL control. The experimental results in the form of graphics and tables have shown the effectiveness of the proposed control algorithm.

@article{mohanty_2020,
title = {3 DOF Autonomous Control Analysis of an Quadcopter Using Artificial Neural Network},
author = {Mohanty S.; Misra A. },
booktitle = {Modern Approaches in Machine Learning and Cognitive Science: A Walkthrough.},
year = {2020},
institution = {Defense Institute of Advanced TechnologyPuneIndia},
abstract = {The Quadcopter is an Unmanned Aerial Vehicle (UAV) which has turned out to be exceptionally mainstream among specialists in the recent past due to the advantages it offers over conventional helicopters. Quadcopter is extremely unique and interesting, however it is inherently unsteady from streamlined features perspective and aerodynamics point of view. In recent past scientists have proposed many control schemes for the stability of quadcopter, but Artificial Neural Network (ANN) systems provide us with the fusion of human intelligence, logic and reasoning. The research focuses on the use of ANN for the control plant systems whose plant dynamics are expensive to model, inaccurate or change with time and environment. In this paper, we explore the Linear Quadratic Regulator (LQR) and Sliding Mode Control (SMC) control is designed for an quadcopter with 3 Degree Of freedom (DOF) Hover model by Quancer. The main benefits of this approach are the model’s ability of adapt quickly to unmodeled aerodynamics, disturbances, component failure due to battle damage, etc. It eliminates the costs and time associated with the wind tunnel testing and generation of control derivatives for the UAV’s. },
keywords = {Quadcopter, Unmanned aerial vehicle (UAV), Artificial neural network (ANN), Linear quadratic regulator (LQR), Sliding mode control (SMC)},
language = {English},
series = {Studies in Computational Intelligence},
publisher = {Springer, Cham},
isbn = {978-3-030-38444-9},
pages = {39-57}
}

The Quadcopter is an Unmanned Aerial Vehicle (UAV) which has turned out to be exceptionally mainstream among specialists in the recent past due to the advantages it offers over conventional helicopters. Quadcopter is extremely unique and interesting, however it is inherently unsteady from streamlined features perspective and aerodynamics point of view. In recent past scientists have proposed many control schemes for the stability of quadcopter, but Artificial Neural Network (ANN) systems provide us with the fusion of human intelligence, logic and reasoning. The research focuses on the use of ANN for the control plant systems whose plant dynamics are expensive to model, inaccurate or change with time and environment. In this paper, we explore the Linear Quadratic Regulator (LQR) and Sliding Mode Control (SMC) control is designed for an quadcopter with 3 Degree Of freedom (DOF) Hover model by Quancer. The main benefits of this approach are the model’s ability of adapt quickly to unmodeled aerodynamics, disturbances, component failure due to battle damage, etc. It eliminates the costs and time associated with the wind tunnel testing and generation of control derivatives for the UAV’s.

Robotic assistive technologies are increasingly used to enhance the physical rehabilitation of patients who have suffered disorders such as strokes. Not only does it make the lives of disabled and elderly patients easier, but it also improves their body functionalities. Robotic assistive technologies offer people a second chance to overcome challenges that come with their disability. The objective of the thesis is to design, prototype and evaluate a 3 Degrees of Freedom (DOF) pneumatic manipulandum for wrist rehabilitation that is capable of accommodating to wrist motions (ulnar deviation, radial deviation, flexion or extension). Since the wrist is the most mobile part of the hand, its post-stroke rehabilitation is difficult. In order to accommodate the wrist motion, 3 DOF are needed. 2 DOF are needed for the horizontal motion and another DOF to allow the manipulandum to move up and down with the wrist. Each DOF is actuated by one pneumatic actuator. The design is prototyped using a 3D printer. The workspace and the required force are analyzed and calculated based on the kinematics of the manipulandum. The pneumatic actuators that were chosen are available in non-magnetic material, which means they are compatible with Functional Magnetic Resonance Imaging (fMRI-compatible). The manipulandum is connected to a Neuro Function Evaluation (NFE) game which is used in the Rehabilitation Centre in Winnipeg. While running the game, the manipulandum is tested and evaluated in assistive and resistive modes. The performance of the manipulandum is analyzed using two methods: image processing and file streaming. The image processing method determines the location of the ball and the location of the paddle of the NFE game in the screen by taking screenshots, while the file streaming method is used to obtain those two locations from the code of the game itself.

@conference{ortega-vidal_2020,
title = {A comparison between optimal LQR control and LQR predictive control of a planar robot of 2DOF},
author = {Ortega–Vidal, A.; Salazar–Vasquez, F.; Rojas–Moren, A.},
booktitle = {2020 IEEE XXVII International Conference on Electronics, Electrical Engineering and Computing (INTERCON)},
year = {2020},
institution = {Universidad de Ingenieria y Tecnologia (UTEC), Peru},
abstract = {This work employs a LQR (Linear Quadratic Regulator) predictive control as well as a LQR controller to control the angular servomotor positions of a planar robot of 2DOF (2 Degrees of Freedom). The goal of such a pantograph type robot is to manipulate the X–Y positions of a 4–bar linkage end effector using two rotary servo base units connected to two revolute joints. Three unactuated revolute joints complete the five links of the robot. Experimental results demonstrate that the LQR predictive controller performs better than the LQR controller, because the former controller is able to diminish the steady–state error between the Cartesian coordinates with respect to the desired coordinates. },
keywords = {LQR controller, LQR predictive controller, inverse kinematics, direct kinematics, planar robot of 2DOF},
language = {English},
publisher = {IEEE},
isbn = {978-1-7281-9378-6}
}

This work employs a LQR (Linear Quadratic Regulator) predictive control as well as a LQR controller to control the angular servomotor positions of a planar robot of 2DOF (2 Degrees of Freedom). The goal of such a pantograph type robot is to manipulate the X–Y positions of a 4–bar linkage end effector using two rotary servo base units connected to two revolute joints. Three unactuated revolute joints complete the five links of the robot. Experimental results demonstrate that the LQR predictive controller performs better than the LQR controller, because the former controller is able to diminish the steady–state error between the Cartesian coordinates with respect to the desired coordinates.

This paper describes a novel stereo vision sensor based on deep neural networks, that can be used to produce a feedback signal for visual servoing in unmanned aerial vehicles such as drones. Two deep convolutional neural networks attached to the stereo camera in the drone are trained to detect wind turbines in images and stereo triangulation is used to calculate the distance from a wind turbine to the drone. Our experimental results show that the sensor produces data accurate enough to be used for servoing, even in the presence of noise generated when the drone is not being completely stable. Our results also show that appropriate filtering of the signals is needed and that to produce correct results, it is very important to keep the wind turbine within the field of vision of both cameras, so that both deep neural networks could detect it.

@conference{lutter_2020,
title = {A Differentiable Newton Euler Algorithm for Multi-body Model Learning},
author = {Lutter, M.; Silberbauer, J.; Watson, J.; Peters, J.},
booktitle = {International Conference on Machine Learning (ICML)},
year = {2020},
institution = {Technical University of Darmstadt, Germany},
abstract = {In this work, we examine a spectrum of hybrid models for the domain of multibody robot dynamics. We motivate a computation graph architecture that embodies the Newton Euler equations, emphasising the utility of the Lie Algebra form in translating the dynamical geometry into an efficient computational structure for learning (Handa et al., 2016). We describe the used actuator models (Section 3) and the virtual parameters (Appendix). In the experiments, we evaluate 26 Newton-Euler based system identification approaches and benchmark these models on the simulated and physical Furuta Pendulum and Cartpole. The comparison shows that the kinematic parameters, required by previous Newton-Euler methods (Atkeson et al., 1986; Sutanto et al., 2020; Ledezma & Haddadin, 2017), can be accurately inferred from data. Furthermore, we highlight that models with guaranteed bounded energy of the uncontrolled system generate non-divergent trajectories, while more general models have no such guarantee. Therefore, their performance strongly depends on the data distribution. The main contributions of this work are the introduction of a white-box model that jointly learns dynamic and kinematics parameters and can be combined with black-box components. We then provide an extensive empirical evaluation on challenging systems and different datasets that elucidates the comparative performance of our grey-box architecture with comparable white and black-box models. },
language = {English}
}

In this work, we examine a spectrum of hybrid models for the domain of multibody robot dynamics. We motivate a computation graph architecture that embodies the Newton Euler equations, emphasising the utility of the Lie Algebra form in translating the dynamical geometry into an efficient computational structure for learning (Handa et al., 2016). We describe the used actuator models (Section 3) and the virtual parameters (Appendix). In the experiments, we evaluate 26 Newton-Euler based system identification approaches and benchmark these models on the simulated and physical Furuta Pendulum and Cartpole. The comparison shows that the kinematic parameters, required by previous Newton-Euler methods (Atkeson et al., 1986; Sutanto et al., 2020; Ledezma & Haddadin, 2017), can be accurately inferred from data. Furthermore, we highlight that models with guaranteed bounded energy of the uncontrolled system generate non-divergent trajectories, while more general models have no such guarantee. Therefore, their performance strongly depends on the data distribution. The main contributions of this work are the introduction of a white-box model that jointly learns dynamic and kinematics parameters and can be combined with black-box components. We then provide an extensive empirical evaluation on challenging systems and different datasets that elucidates the comparative performance of our grey-box architecture with comparable white and black-box models.

@article{wang3_2020,
title = {A dual adaptive fault-tolerant control for a quadrotor helicopter against actuator faults and model uncertainties without overestimation},
author = {Wang, B.; Yu, X.; Mu, L.; Zhang, Y.},
journal = {Aerospace Science and Technology},
year = {2020},
month = {4},
volume = {99},
institution = {Northwestern Polytechnical University, China; Beihang University, China; Xi'an University of Technology, China, Concordia University, Canada},
abstract = {This paper presents a dual adaptive fault-tolerant control strategy for a quadrotor helicopter based on adaptive sliding mode control and adaptive boundary layer. Within the proposed adaptive control strategy, both model uncertainties and actuator faults can be compensated without the knowledge of the uncertainty bounds and fault information. By virtue of the proposed adaptive control scheme, the minimum discontinuous control gain is adopted, which significantly reduces the control chattering effect. As compared to the existing adaptive sliding mode control schemes in the literature, larger actuator faults can be tolerated by employing the proposed control scheme while suppressing control chattering. Moreover, boundary layer is used to smoothen control discontinuity and further eliminate control chattering. Nevertheless, the choice of boundary layer thickness is a trade-off between system stability and tracking accuracy. By explicitly considering this fact, an adaptive boundary layer is developed and synthesized with the proposed adaptive control framework to ensure stability and tracking accuracy of the considered system. When the control parameter tends to be overestimated, the thickness of boundary layer can be appropriately adjusted to avoid control parameter overestimation. Simulation and experimental tests of a quadrotor helicopter are both conducted to validate the effectiveness of the proposed control scheme. Its advantages are demonstrated in comparison with a conventional adaptive sliding mode control scheme. },
keywords = {Actuator fault, Adaptive sliding mode control, Fault-tolerant control, Model uncertainty, Quadrotor helicopter},
language = {English},
publisher = {Elsevier Masson}
}

This paper presents a dual adaptive fault-tolerant control strategy for a quadrotor helicopter based on adaptive sliding mode control and adaptive boundary layer. Within the proposed adaptive control strategy, both model uncertainties and actuator faults can be compensated without the knowledge of the uncertainty bounds and fault information. By virtue of the proposed adaptive control scheme, the minimum discontinuous control gain is adopted, which significantly reduces the control chattering effect. As compared to the existing adaptive sliding mode control schemes in the literature, larger actuator faults can be tolerated by employing the proposed control scheme while suppressing control chattering. Moreover, boundary layer is used to smoothen control discontinuity and further eliminate control chattering. Nevertheless, the choice of boundary layer thickness is a trade-off between system stability and tracking accuracy. By explicitly considering this fact, an adaptive boundary layer is developed and synthesized with the proposed adaptive control framework to ensure stability and tracking accuracy of the considered system. When the control parameter tends to be overestimated, the thickness of boundary layer can be appropriately adjusted to avoid control parameter overestimation. Simulation and experimental tests of a quadrotor helicopter are both conducted to validate the effectiveness of the proposed control scheme. Its advantages are demonstrated in comparison with a conventional adaptive sliding mode control scheme.

@article{reyes-baez_2020,
title = {A family of virtual contraction based controllers for tracking of flexible-joints port-Hamiltonian robots: theory and experiments},
author = {Reyes-Baez, R.; van der Schaft, A.; Jayawardhana, B.; Pan, l.},
journal = {arXiv},
year = {2020},
institution = {University of Groningen, The Netherlands},
abstract = {In this work we present a constructive method to design a family of virtual contraction based controllers that solve the standard trajectory tracking problem of flexible-joint robots (FJRs) in the port-Hamiltonian (pH) framework. The proposed design method, called virtual contraction based control (v-CBC), combines the concepts of virtual control systems and contraction analysis. It is shown that under potential energy matching conditions, the closed-loop virtual system is contractive and exponential convergence to a predefined trajectory is guaranteed. Moreover, the closed-loop virtual system exhibits properties such as structure preservation, differential passivity and the existence of (incrementally) passive maps. },
keywords = {Flexible-joints robots, tracking control, port-Hamiltonian systems, contraction, virtual control systems},
language = {English}
}

In this work we present a constructive method to design a family of virtual contraction based controllers that solve the standard trajectory tracking problem of flexible-joint robots (FJRs) in the port-Hamiltonian (pH) framework. The proposed design method, called virtual contraction based control (v-CBC), combines the concepts of virtual control systems and contraction analysis. It is shown that under potential energy matching conditions, the closed-loop virtual system is contractive and exponential convergence to a predefined trajectory is guaranteed. Moreover, the closed-loop virtual system exhibits properties such as structure preservation, differential passivity and the existence of (incrementally) passive maps.