Final Report Summary - H2R (Bringing Human Neuromotor Intelligence to Robots) In the near future, technologies will be developed to enable robots working in unknown dynamic environments, and to physically interact with human beings in a safe and friendly manner. Consequently, future personal robots could be integrated seamlessly into our daily life in the fields of service, health care, entertainment, etc. However, most robot control technologies nowadays are developed for industrial purposes and fail at gently interacting with humans, and are not able to cope with the varying tasks of daily situations, which cannot be predefined. On the contrary, our human neuromotor control systems perform much better in terms of versatility, dexterity and safety. Therefore, the objective of this project is to bring new ideas and strategies extracted from human motor control system to robot control systems. To be more specific, this project will use advanced control theory and machine learning techniques to model the main properties of the unique human neuro-muscular control system, and then develop human-like adaptive learning controllers based on the modeling of human neuromotor controller.This project first has investigated human's adaptation of force and impedance in interactive tasks by means of modelling and analysis. The adaptation algorithms have been then modified to apply unto exoskeleton robots, which are developed to support human rehabilitation process. The algorithms have been also applied on iCub robot for bimanual manipulation tasks. Secondly, the control is investigated for over actuated systems and under actuated systems, in which the number of actuators is less than the number of degree of freedoms (DOFs). Inspired by the way that human operator controls dynamic balanced two-wheel vehicles, an adaptive generator of indirect control trajectories has been developed to manipulate the passive DOFs' motion by using the dynamic coupling with active DOFs. The developed method has been used to control wheeled pendulum system as well manipulators of up to a half passive joints. Thirdly, because adaptation of reference trajectories have been observed in human motor experiments, this project has developed computational model to reveal experimental results. The developed algorithms have been also applied for robotic haptic identification without using force sensors. Fourthly, the project studies how human learn proper impedance to perform various tasks, and thereafter developed an optimal impedance model reference control using linear quadratic regulator optimization techniques. The project also carried out experimental study of how to achieve best kicking performance using suitable compliance in robot leg joints. Lastly, this project studies Electromyography (EMG) signals, which can reflect the muscle activation. More robust boosting algorithm based pattern recognition method has been proposed, and force/impedance estimation using EMG has been investigated. Motion/force control method using EMG has been developed and tested on exoskeleton robots and robot manipulators. So far, the research of this project has generated a number of novel robot controllers that apply to assistive rehabilitation, human-robot cooperation, human-like robot automatic control, robot haptic identification et al. And these control algorithms have been implemented and tested on exoskeleton robots, humanoid robots and robot manipulators. These results have provided new means of biomimetic robot control technologies to both academic and industrial communities. These results are expected to deeply impact future control technologies of robot applications where successful task completion requires people and robots to collaborate directly in a shared workspace or robots to move autonomously and safely. The developed control algorithms are expected to be employed in future personal robots, the market of which is predicted to be worth $15 billion by 2015.The knowledge created by this project has been well transferred to the University of Plymouth through numerous seminars, talks, research meetings, discussions with students and colleagues, as well as internal reports. The knowledge has also been transferred to Imperial College London, via collaboration with Professor Etienne Burdet, through numerous in person and Skype discussions, reports and talks. In addition, knowledge transfer to other institutes in Europe and China has been achieved through academic visits, exchanges, talks, and collaborations with University of Portsmouth, University of Glamorgan, University of Bristol, Bristol Robotics Laboratories, University of Essex, Coventry University, Brunel University, The Chinese University of Hong Kong, South China University of Technology, Northwestern Polytechnical University, Xi'an Jiaotong University, Xidian University, and Harbin Institute of Technology.