International Journal of Control Automation and Systems最新文献

With the increasing complexity of the robot grasping environment, it puts forward higher requirements on the grasping strategy of manipulators. However, grasping cluttered multi-class objects is a challenging task because the objects are stacked and occluded from each other, and it is often difficult for robots to find a suitable grasping position. Deep reinforcement learning with DQN algorithm has been used to study the pushing and grasping strategy to manipulate cluttered multi-class objects. However, there exists long learning time and low success rate. To solve this problem, we adopt two fully convolutional networks (FCN) to map the color and depth maps to actions: pushing and grasping. These two networks are trained by an improved soft actor-critic algorithm that includes auto entropy regularization, regularized objective function and clipped double Q learning. Pushing and grasping synergies has been learned with the dense reward feedback. The simulation experiment demonstrate that the learning process converges quickly and stable with a grasp success rate up to 83.3%. We further demonstrate the generalization ability and well performance of our models when novel objects appear in the scenes that the robot has never grasped before. Finally, real-world experiments with trained models from simulations are conducted to test the grasping performance on manually arranged scenes.

This paper investigates the N-coalition game, which covers the generalized Nash equilibrium problems (GNEPs) and distributed optimization problems. To seek the Nash equilibrium of the N-coalition games, we firstly exploit a centralized algorithm. Under the full information, the algorithm can exponentially converge to the Nash equilibrium of the N-coalition games. Then, based on the extremum seeking (ES) approach, we extend the centralized algorithm to distributed counterpart. Different from the centralized algorithm, the explicit cost functions are not available in the distributed algorithm. The players not only can converge to a small neighborhood of the Nash equilibrium, but do not communicate with the other coalitions. Furthermore, in order to reduce the communication burden in intra-coalition, an event-triggered based distributed algorithm is proposed. By the algorithm, the players only communicate with their neighbors in intra-coalition at event-triggered time constants, and also can converge to a neighborhood of the Nash equilibrium of the N-coalition games. Finally, an example about Nash-Cournot game is given to illustrate the effectiveness of our algorithms.

This article mainly tackles the finite-time time-varying formation tracking (Fin-TVFT) problem and fixed-time time-varying formation tracking (Fix-TVFT) problem of multiple Euler-Lagrange systems (MELSs) subject to the external disturbances under the directed interaction topology. The distributed estimator-based hierarchical control (EBHC) algorithms are respectively formulated to achieve the foregoing problems. Note that the settling time of Fin-TVFT depends on the initial values, but the convergence time of Fix-TVFT is independent of the initial conditions. According to the Lyapunov stability analysis, several sufficient conditions for accomplishing the Fin-TVFT and Fix-TVFT are derived. Eventually, the simulation results are exhibited in order to illustrate the feasibility of the designed algorithms.

This article presents a data-driven control application to robot manipulation for implementing the time-delayed control (TDC) algorithm. TDC scheme uses the previous information to cancel out all the dynamics except the inertial torque in robot manipulators. The accuracy of estimating the inertia matrix plays an important role in control performance as well as the stability of TDC. Necessary information for the time-delayed control is inertia and acceleration signals. Since selecting the constant inertia matrix is simple but concerned with the poor performance, better estimation is required. Based on the input and output data of a robot manipulator, necessary models are obtained by a recursive least squares (RLS) algorithm and those models are used for estimating acceleration signals by designing a state observer (SOB). Here the models of a robot arm are decoupled, linearized, and identified by RLS algorithm and the joint acceleration signals are identified by a state observer in on-line fashion. Combining RLS, SOB, and TDC yields RST scheme for a robot manipulator to improve the tracking control performance by providing solutions for TDC problems. Tracking control performances of a mobile manipulator by the RST scheme are empirically tested.

In this paper, the integral event-triggered time-varying formation tracking (TVFT) problem is investigated for unmanned aerial vehicle (UAV) swarm systems with switching directed topologies. An improved integral event-triggered mechanism (ETM) is introduced into the TVFT control protocol, which can further reduce communication frequency and resource consumption compared with traditional static ETM. The switching topology is considered to improve the reliability of the communication network. Different from traditional TVFT, the addition of the switching signal and the ETM brings challenges to the system analysis and design. To this end, the control protocol of TVFT, the admissible edge-dependent average dwell time switching signal and the ETM are co-designed, the global exponential stability criterion is further deduced. The simulation verifies the effectiveness and merits of the design scheme.

As unmanned aerial vehicles (UAVs) have limited energy resources and diverse constraints, the reconfiguration of their formation encounters substantial challenges. In this paper, we employ a leader-follower method. In order to minimize flight distance and resource consumption, a greedy algorithm is used to allocate leader and follower positions. Based on the limitations of the receding horizon control (RHC) method and the control parameterization and time discretization (CPTD) method, we propose the no reference path RHC (NRPRHC) method. The proposed method transforms the formation reconfiguration into smaller local optimization problems, leading to a reduction in the size of the optimization stages and computational complexity. For each local optimization problem, we propose the adaptive population differential evolution (APDE) algorithm to optimize the control inputs. Finally, the results are provided to illustrate the feasibility and effectiveness of the proposed method.

In response to the increasingly complex problem of patrolling urban areas, the utilization of deep reinforcement learning algorithms for autonomous unmanned aerial vehicle (UAV) coverage path planning (CPP) has gradually become a research hotspot. CPP’s solution needs to consider several complex factors, including landing area, target area coverage and limited battery capacity. Consequently, based on incomplete environmental information, policy learned by sample inefficient deep reinforcement learning algorithms are prone to getting trapped in local optima. To enhance the quality of experience data, a novel reward is proposed to guide UAVs in efficiently traversing the target area under battery limitations. Subsequently, to improve the sample efficiency of deep reinforcement learning algorithms, this paper introduces a novel dynamic soft update method, incorporates the prioritized experience replay mechanism, and presents an improved deep double Q-network (IDDQN) algorithm. Finally, simulation experiments conducted on two different grid maps demonstrate that IDDQN outperforms DDQN significantly. Our method simultaneously enhances the algorithm’s sample efficiency and safety performance, thereby enabling UAVs to cover a larger number of target areas.

In this paper, the optimal backstepping control method based on game theory is designed for strict feedback nonlinear systems with input saturation. In order to improve the robustness of the system, the zero-sum game problem of control input and disturbance for strict feedback systems is studied in this paper. Specifically, reinforcement learning (RL) is used to obtain the Nash equilibrium solution of the subsystem corresponding to the virtual control input and the virtual disturbance based on the HJB equation. By using the recursive process of backstepping, the controller of the tracking game problem between the control input and the disturbance of the high-order system is designed. In addition, according to the Lyapunov stability theory, it is proved that all internal signals of the closed-loop systems are uniformly ultimately bounded (UUB). Finally, simulation results are provided to illustrate the validity of the proposed method.

The main purpose of this paper is to study the stability of discrete-time impulsive switched T-S fuzzy systems with two kinds of asynchronous behaviors, including asynchronous behavior between impulse and switching, and asynchronous switching between controllers and subsystems. We divide the subsystems into stable and unstable subsystems, which respectively adopt slow switching and fast switching methods. Then, based on multiple Lyapunov functions, admissible edge-dependent average dwell time (AED-ADT) and admissible edge-dependent average impulsive interval (AED-AII) methods, sufficient conditions for global uniform exponential stability (GUES) of the closed-loop system are established, and the results are less conservative than that based on mode-dependent average dwell time (MDADT) and mode-dependent average impulsive interval (MDAII) methods. In addition, we provide the solvability conditions for the state feedback controller. Finally, several numerical examples are provided to verify the effectiveness of the results in this paper.

This paper presents a complete system architecture for multi-robot coordination for unbalanced task assignments, where a number of robots are supposed to visit and accomplish missions at different locations. The proposed method first clusters tasks into clusters according to the number of robots, then the assignment is done in the form of one-cluster-to-one-robot, followed by solving the traveling salesman problem (TSP) to determine the visiting order of tasks within each cluster. A nonlinear model predictive controller (NMPC) is designed for robots to navigate to their assigned tasks while avoiding colliding with other robots. Several simulations are conducted to evaluate the feasibility of the proposed architecture. Video examples of the simulations can be viewed at https://youtu.be/5C7zTnv2sfo and https://youtu.be/-JtSg5V2fTI?si=7PfzZbleOOsRdzRd. Besides, we compare the cluster-based assignment with a simulated annealing (SA) algorithm, one of the typical solutions for the multiple traveling salesman problem (mTSP), and the result reveals that with a similar optimization effect, the cluster-based assignment demonstrates a notable reduction in computation time. This efficiency becomes increasingly pronounced as the task-to-agent ratio grows.