Man-machine interface prptotype for realtime prediction of motion of exoskeleton's users

In the United States, over 282.000 people today are affected by Spinal Cord Injury, which is a traumatic damage to the spinal cord that leads to a partial or total body paralysis. At the UC Berkeley Human Robotics & Engineering Laboratory has been developed the Phoenix Exoskeleton which aims to reduce the impediments due to this physical damage by replacing the wheelchair in order to give to the patients the opportunity of changing their quality life, by improving overall health and mobility. This work introduces an innovative approach for replacing the current user interface, allowing to the user to have an interaction with the device totally free of active inputs, while walking, through a safe system which predicts in real time the users intentions. The software prototype is designed for providing an MMI (Man-Machine Interface) able to reduce the time gap and the balance variation between the moment when the intention of walking presents itself and the trigger of the swing really happens, according to the empiric hypothesis formulated by the author which assumes that exoskeletons users have involuntary body movements which differentiate their motion desires. The system has been developed using a machine learning model fed with post- processed data collected by two encoders and one IMU, which describe the users gait. It has been called Body Motion Triggering. The efficiency of the prototype has been tested in laboratory by the pilot: the system turned out to be totally safe; the predicting outcomes still need to be widely improved but the achieved results are encouraging and have demonstrated the validity of the formulated hypothesis, opening a new potential field of research. The hope behind the analysis carried out is to have been useful in order to improve the lives of people that need it.