Flexible effects in the joints of large space robots make their real-time operation a challenging task, especially when accurate endpoint positioning is required. The problem is further aggravated when the flexible joint stiffness matrix is not well known. This paper discusses the application of a model reference adaptive control (MRAC) composite system for tracking the endpoint of a flexible joint space robotic manipulator. The composite control scheme consists in a flexible control term designed to damp the joint vibrations plus a Transpose Jacobian rigid control term for which the control gains are adapted using a novel direct MRAC adaptation law. Numerical simulations show that the adaptive composite controller can maintain stability and good tracking performance despite significant uncertainties in the joint stiffness coefficients.
Nomenclaturecomponent of the adaptive derivative gain matrix ) (t dp K = proportional component of the adaptive derivative gain matrix i k = stiffness of joint i , n i , , 1 K = p K = constant proportional gain matrix ) (t p K = adaptive proportional gain matrix ) (t pi K = integral component of the adaptive proportional gain matrix ) (t pp K = proportional component of the adaptive proportional gain matrix v K = vibration damping gain matrix ci l = distance from previous joint to the center of gravity of link i , n i , , 1 K = i l = length of link i , n i , , 1 K = i m = mass of link i , n i , , 1 K = ) (q M r = rigid inertia matrix 2 q = link angles vector m q = motor angles vector x , y = actual endpoint position c x , c y = commanded endpoint position ref x , ref y = model reference endpoint position p δ , d δ = small control coefficients pp Γ , pi Γ = proportional control parameters dp Γ , di Γ = derivative control parameters τ = control torque vector f τ = fast control torque vector s τ = slow control torque vector