Control of constrained manipulators with flexible joints (Q1911408)

For robots with elasticity concentrated at the joints having the end-effector constrained to a cartesian surface, the paper proposes a simple PD controller plus constant feedforward for the task of regulating both the end-effector position and the normal force exerted on the surface. The proportional-derivative action is based only on position and velocity measures on the motor side of the elastic joints. The feedforward is the nominal gravity compensation at the desired configuration and the associated nominal joint-space torque associated to the desired cartesian force. A sufficient condition on the PD gains is proved under which global regulation is achieved. The main limitation of the approach is that a bilateral constraint is used to model the interaction with the environment. This implies that the end-effector is always in contact with the environment, so that the force multipliers associated to the constraint are allowed to assume arbitrary positive or negative values during the transient response of the closed-loop system. In reality, constraints are unilateral and force multpliers with constant (positive) sign should be enforced at all times for not losing contact. The proposed controller does not guarantee this property.