Final Report Summary - H2R (Bringing Human Neuromotor Intelligence to Robots) In the previous phase, our research study focuses on the development of novel robot control strategies inspired by human neuromotor control principles, especially on human-like control design for a single robot arm. In this phase, we aim to extend our previous biomimetic control methods designed for single arm robot system to multiple robot arms system, especially on dual-arm robot system. As we know, bimanual manipulation is perhaps the most widely employed approach in the nature, especially on our primates. We also aim to apply our previous research of human-robot interaction from in face operation to teleoperation. As we know, telerobots have a wide range of application in space and subsea exploration as well as to manipulate dangerous and hazardous objects, e.g. disabling a bomb. Nowadays, telerobot are also coming into medical fields, e.g. tele-operated microsurgery, and health care, e.g. tele-assisted rehabilitation. In this project, we have first investigated force and impedance adaptation in dual-arm coordinated control. A task/joint space hybrid adaptive mechanism has been developed, to counter the external disturbance applied at either the end effectors or other parts of the arm. In addition, we have introduced a fuzzy logic based inference system to select most suitable tuning gains in the adaptation algorithm. Furthermore, we have also attempted to study dual-arm and multiple-arm coordination in the framework of multi-agent system. For this purpose, effort has been made on theoretic study of uncertain couplings in multi-agent system. On the other hand, we have investigated fuzzy based adaptive control and hand gesture based control in teleoperation using humanoid robots, e.g. iCub robot and Baxter robot, based on our modeling of these robots. Two haptic devices, Phantom Omni and Falcon have been employed to implement force feedback. Various techniques in pattern recognition and adaptive tuning have been investigated, in order to achieve best possible teleoperation performance. Numerous research outputs have been generated out of this project, and these results have provided us a solid base to further our study on robot applications, especially on teleoperation. As the combination of a local human operator and a remote robot, the telerobot system should merge the intelligence of both robot and operator to achieve an optimal performance. We hypothesize that best performance would be achieved when the human operator motion pattern and the telerobot controller pattern are perfectly matched. In the future, we expect to use adaptive dynamic programming (ADP) enhanced neural networks (NNs) technique to capture human skills, and to tune the telerobot controller for optimal matching with human operator's skills. For this purpose, effort has also been made on ADP based NN control theory.The knowledge created by this project has been well transferred to the Beijing Institute of Technology through seminars, talks, research meetings, discussions with students and colleagues. Through collaboration, the knowledge has also been transferred to other Chinese institutes including the Chinese University of Hong Kong, South China University of Technology, Harbin Institute of Technology, Institute of Automation at Chinese Academy of Sciences, Northwestern Polytechnical University, Liaoning University of Technology and Xidian University, and universities in the UK, including University of Plymouth, Imperial College London, University of Portsmouth, and University of Bristol.