Position/force control of biped walking robots

This paper addresses the problem of modelling and control of biped robot by combining Cartesian-based position and force control algorithms. The complete walking cycle is divided in two phases: i) single support - in which it is studied the trajectory controllability based on simple motion goals and ii) exchange of support - in which the foward leg absorbs the impact and then gradually accept the robot's weight. The control of the foot with the constrained surface is modelled through linear spring-damper systems. The system's controllability is enhanced through the insertion of a dynamic selection matrix that modifies the actuating profile in each phasse. The control algorithms are simulated and their effectiveness and robustness are discussed.