Adaptive position control of fluidic soft-robots working with unknown loads

The compliance of robot structure is essential to provide safety if robots are working in the direct contact with humans. Soft-actuators possess natural inherent compliance but they are characterized with highly non-linear dynamics making accurate control a challenging task. This paper presents the robust and adaptive position control of soft-robots based on pneumatic actuators with rotary elastic chambers (REC-actuators). Previous work shows that accurate control of soft-robots is possible if load parameters are known and taken into account in quasi-static model of soft-robots, that is used as feed-forward signal. This paper demonstrates that adaptation of the robot model provides flexibility and robustness when soft-robots deal with unknown loads. It is shown that due to adaptation the load parameters are not needed to be changed in the quasi-static robot model. The performance of the adaptive controller is proven in simulation and in experiments using firstly a simple 1 degree of freedom (DoF) and then a complex 6 DoF robot structure.