When a mobile robot is required to perform tasks in the unknown and complex environment, it is critical to have the ability of dynamic obstacle avoidance. However, conventional deep deterministic policy gradient (DDPG) for collision-free navigation can only perceive a fixed number of dynamic obstacles, and thus it cannot adapt to the stochastic working scenario. To overcome the limitation, an improved DDPG algorithm is proposed in this study. It is an exploration to implement the DDPG with long short-term memory (LSTM) network-based encoder to achieve dynamic obstacle avoidance for the mobile robot in the stochastic working scenario, which can encode the variable number of obstacles into a fixed-length representation. Specifically, to facilitate the LSTM network-based encoder, one safe processing rule is designed to guarantee the entire information of the observable obstacles can be represented completely. The LSTM network-based encoder takes the latest environment information of observable obstacles by employing the safe processing rule and generates the fixed length state vector. In addition, continuous state space for mobile robot and obstacles, as well as reward function and action space are designed. Both simulations and experiments are conducted and the results verify that the improved DDPG algorithm can achieve collision-free trajectory with multiple dynamic obstacles well. It helps to reduce the path distance and motion time effectively.

Human–robot-collaboration requires robot to proactively and intelligently recognize the intention of human operator. Despite deep learning approaches have achieved certain results in performing feature learning and long-term temporal dependencies modeling, the motion prediction is still not desirable enough, which unavoidably compromises the accomplishment of tasks. Therefore, a hybrid recurrent neural network architecture is proposed for intention recognition to conduct the assembly tasks cooperatively. Specifically, the improved LSTM (ILSTM) and improved Bi-LSTM (IBi-LSTM) networks are first explored with state activation function and gate activation function to improve the network performance. The employment of the IBi-LSTM unit in the first layers of the hybrid architecture helps to learn the features effectively and fully from complex sequential data, and the LSTM-based cell in the last layer contributes to capturing the forward dependency. This hybrid network architecture can improve the prediction performance of intention recognition effectively. One experimental platform with the UR5 collaborative robot and human motion capture device is set up to test the performance of the proposed method. One filter, that is, the quartile-based amplitude limiting algorithm in sliding window, is designed to deal with the abnormal data of the spatiotemporal data, and thus, to improve the accuracy of network training and testing. The experimental results show that the hybrid network can predict the motion of human operator more precisely in collaborative workspace, compared with some representative deep learning methods.

When the mobile robot performs certain motion tasks in complex environment, wheel slipping inevitably occurs due to the wet or icy road and other reasons, thus directly influences the motion control accuracy. To address unknown wheel longitudinal slipping problem for mobile robot, a RBF neural network approach based on whole model approximation is presented. The real-time data acquisition of inertial measure unit (IMU), encoders and other sensors is employed to get the mobile robot’s position and orientation in the movement, which is applied to compensate the unknown bounds of the longitudinal slipping using the adaptive technique. Both the simulation and experimental results prove that the control scheme possesses good practical performance and realize the motion control with unknown longitudinal slipping.

Magnetic wall-climbing robots have wide applications due to their large adhesive force and loading capacity. However, researches have seldom focused on the magnet patterns and magnetic characteristics of the adhesive module, which may influence the output performance of the robot greatly. Therefore, the purpose of this study is to analyze the magnet array, flux distribution and magnetic adhesive force of wall-climbing robot in both quantitative and qualitative ways. It helps to improve the magnetic force and loading capacity significantly. The concept design of the wall-climbing robot is introduced and the system stability is analyzed. Following that, four magnet patterns are presented and the magnetic flux distribution is numerically simulated. The comparison among them shows that the alternatively magnetization pattern can generate larger magnetic flux density relatively. Subsequently, the magnetic field and adhesive force are both formulated analytically. It is worth noting that the curved surface is considered for the analytical modeling in this study. Thus, the modeling approach can be implemented to the magnetic characteristic analysis of other similar magnetic adhesive robots. Then, the numerical simulation is conducted on the magnetic field and force output to validate the analytical models. One research prototype of the magnetic wall-climbing robot is developed, and the system hardware architecture is introduced. Experiments are carried out to validate the analytical model and output performance of the robot system. The experimental result is consistent with the analytical model and numerical computation.

When the wheeled mobile robot (WMR) is required to perform specific tasks in complex environment, i.e., on the forestry, wet, icy ground or on the sharp corner, wheel skidding and slipping inevitably occur during trajectory tracking. To improve the trajectory tracking performance of WMR under unknown skidding and slipping condition, an adaptive sliding mode controller (ASMC) design approach based on the extended state observer (ESO) is presented. The skidding and slipping is regarded as external disturbance. In this paper, the ESO is introduced to estimate the lumped disturbance containing the unknown skidding and slipping, parameter variation, parameter uncertainties, etc. By designing a sliding surface based on the disturbance estimation, an adaptive sliding mode tracking control strategy is developed to attenuate the lumped disturbance. Simulation results show that higher precision tracking and better disturbance rejection of ESO-ASMC is realized for linear and circular trajectory than the ASMC scheme. Besides, experimental results indicate the ESO-ASMC scheme is feasible and effective. Therefore, ESO-ASMC scheme can enhance the energy efficiency for the differentially driven WMR under unknown skidding and slipping condition.

When the robot is required to execute a certain task in the world coordinate system (WCS), it is necessary to find the coordinate transformation between the robot base coordinate system (RBCS) and WCS to enable the high precision motion planning. This paper presents a simple and accurate method that allows a simultaneous computation of the coordinate transformations (i.e., rotation and translation) from WCS to RBCS. Based on the dual quaternion, the robot kinematic model and formulas for calculating the transformation are derived, which allow calculating the rotation and translation simultaneously. Taking the unit dual quaternion as a constraint, the Lagrangian multiplier method is employed to obtain the optimum transformation. Both simulation and experiment results show that higher calibration precision is obtained. The proposed approach has certain reference value and guiding sense for other calibration problems.