Learning variable impedance control is a powerful method to improve the performance of force control. However, current methods typically require too many interactions to achieve good performance. Data-inefficiency has limited these methods to learn force-sensitive tasks in real systems. In order to improve the sampling efficiency and decrease the required interactions during the learning process, this paper develops a data-efficient learning variable impedance control method that enables the industrial robots automatically learn to control the contact force in the unstructured environment. To this end, a Gaussian process model is learned as a faithful proxy of the system, which is then used to predict long-term state evolution for internal simulation, allowing for efficient strategy updates. The effects of model bias are reduced effectively by incorporating model uncertainty into long-term planning. Then the impedance profiles are regulated online according to the learned humanlike impedance strategy. In this way, the flexibility and adaptivity of the system could be enhanced. Both simulated and experimental tests have been performed on an industrial manipulator to verify the performance of the proposed method.

Kinematic parameters' calibration is a powerful method to improve the accuracy of the robot. This paper proposes an effective kinematic self-calibration method for dual-manipulators based on virtual constraints to estimate the actual kinematic parameters of the robots. This method only needs a camera mounted on one robot end-effector (EE) and a calibration target attached to another robot EE. First, a new calibration error model based on the straight line constraint is established to formulate the positions' misalignment error with the kinematic parameters' error. Then, the particle swarm optimization algorithm is developed to generate the optimal calibration poses of the robots under the constraints, which are used to ensure the poses feasible and the measurement errors acceptable. Finally, the kinematic parameter errors are identified with the Levenberg-Marquardt algorithm. The experiments of the kinematic parameters' calibration with the dual-manipulators system are designed. The experimental results showed that the high positional accuracy of both robots can be achieved.

Target recognition and tracking is the key destination in robot vision system. A new efficient algorithm is introduced in this paper to overcome the contradiction among the complexity of algorithm, tracking precision, and the rapidity in real-time system. Firstly, the algorithm is based on the feature of template cursory matching by lowering resolution and using sequential similarity detection algorithm (SSDA) with four items of improvement to locate the position. Secondly, it brings forward a template update algorithm based on confidence level Maximum Close Distance, as the target varying posture and moving constantly. Finally, Kalman filter algorithm is used to estimate the position and it can reduce the ratio of losing target when the target is sheltered. The experiments and simulations show that the algorithms given in the paper have advantages of improving orientation precision, rapidity, practicability, and robustness in orientating target.

The Scale Invariant Feature Transform, SIFT, is invariant to image translation, scaling, rotation, and is partially invariant to illumination changes. But, the time of features extraction and matching is huge, and the number of features is much larger then that is needed. To reduce the number of features generated by SIFT as well as their extraction and matching time, a modified approach based sampling is proposed. Mean-Shift algorithm is used in this modified SIFT to search local extrema points actively in scale space to improve the efficiency. It is demonstrated that the features extracted by modified SIFT are uniformly distributed in space, the time of feature extraction and matching is reduced obviously and the feature matching is accurate.

Compared with the robots, humans can learn to perform various contact tasks in unstructured environments by modulating arm impedance characteristics. In this article, we consider endowing this compliant ability to the industrial robots to effectively learn to perform repetitive force-sensitive tasks. Current learning impedance control methods usually suffer from inefficiency. This paper establishes an efficient variable impedance control method. To improve the learning efficiency, we employ the probabilistic Gaussian process model as the transition dynamics of the system for internal simulation, permitting long-term inference and planning in a Bayesian manner. Then, the optimal impedance regulation strategy is searched using a model-based reinforcement learning algorithm. The effectiveness and efficiency of the proposed method are verified through force control tasks using a 6-DoFs Reinovo industrial manipulator.