Gravity Identification and Admittance Variable Loading Compensation for Single Joint Rehabilitation Exoskeletons

This paper proposes an admittance-like control strategy for gravity identification and variable loading compensation to encourage active patient participation in robotic rehabilitation. An integral backstepping controller (IBC) is used in the position loop of a second admittance model to track the desired reference trajectory of a knee-joint rehabilitation exoskeleton attached to the shank of a clinical mannequin with unknown parameters. A short and seamless online identification phase is used to obtain the gravitational constant by tracking the admittance path generated from a staircase reference with zero steady state error and matching the torque commands to the position outputs when the system is at steady state. The obtained gravitational constant is employed during rehabilitation exercises to vary the loading on the mannequin's knee joint. The effectiveness of the method is tested using a human participant to emulate the application of knee joint torque by turning a shaft of the actuator motor and measuring the interaction torque with a joint torque sensor. The system is implemented in real time on an ESP32 microcontroller interfaced with the actuator and torque sensor using I2C protocol.