Integral sliding mode approaches combined with underlying model-free vs. model-based controllers for a pneumatically actuated assistance system

This paper presents nonlinear control approaches for a prototypical assistance system driven by pneumatic artificial muscles (PAMs) for load handling. A cascaded control structure consisting of two loops is employed to control the internal muscle pressures and the end-effector position. In each loop, two different controller variants are implemented. In the inner cascade, two control strategies are designed and compared with each other. Either a model-free (MF) or a model-based (MB) controller is employed to control the pressure within the PAMs. In addition, two alternative control approaches are implemented in the outer loop: An integral sliding mode (ISM) or an adaptive super-twisting (ASTW) ISM controller stabilizes and maintains the end-effector position despite variations in the payload. After a simulation-based evaluation of both position controllers, the proposed control architecture is successfully validated in experiments at a dedicated test rig.