Design and development of a human-machine interactive-force controlled upper-limb powered exoskeleton for human augmentation and disability rehabilitation

Humanity has long dreamt of synergistically combining human intelligence with robotic strength enhancement. Current technological limitations still pose major challenges to satisfactorily achieving this dream. However, one of the greatest challenges to the creation of a practical, actively-controlled powered exoskeleton (EXOS) lies in its Human-Machine Interface. Brain-Machine Interface EXOS control systems theoretically provide the ultimate in development potential, but they are currently far from achieving satisfactory combinations of safety, practicality, robustness and performance. Hence, an alternative EXOS control paradigm was proposed, using the interactive forces between the EXOS and its user as the input signals. By minimizing these forces, the EXOS should accurately shadow the user’s motions with minimal impedance to one’s motions. Conversely, the level of resistance to the EXOS user’s motions could also be easily controlled for rehabilitative applications. Thus, the main goal of this thesis was to assess the viability of the Human-Machine Interactive-Force Control System (HMIFCS). To provide a realistic basis for the assessment of the HMIFCS’s potential, a Computer Aided Design model of an upper-limb EXOS was created based on existing components and materials. Studies were carried out to assess normal human arm performance to aid in the selection of actuators and sensors with the appropriate performance envelops. After finalization of the EXOS arm’s design, the HMIFCS’s algorithms were designed. A simulation model was created to accurately assess the performance potential of the HMIFCS. The performance constraints of the EXOS arm were included into the HMIFCS simulation model to maximize realism. The effects of various relevant variables (e.g. dynamic compensation, arm length, etc.) on the HMIFCS’s performance were assessed via the simulation model. With the simulation results as the basis, calibrations were made to the HMIFCS’s power assistance settings to optimize free motion performance by minimizing impedance to free motion. However, the free motion optimized settings were incapable of providing safe and robust load attenuation performance. Hence the HMIFCS’s torque amplification curve was modified. The resulting Dynamic Assist Ratio Curve was shown to provide safe and effective strength augmentation without significant affecting free motion performance. A Dynamic Damping Coefficient Curve was also tested and implemented to further refine the HMIFCS’s overall performance. An inherent tendency to significantly attenuate external shocks to the user was also shown, which benefitted safety. Conclusively, the HMIFCS was shown to have excellent potential as a safe, high-performance means to control an EXOS, deserving further research and development via experimentation with physical prototypes.