Compliant control of robot manipulators in environments with unknown or variable dynamics is a broad area of research which has many unresolved problems. Due to the dynamic coupling between the robot and the environment in the constrained directions, hybrid control schemes seem best suited to perform the control since they can be tuned independently in each direction. The directional adaptive damping control scheme is an example of a hybrid control scheme that automatically determines the direction of the constraint and adapts the force response according to the surface stiffness. The extension of this scheme to the more general case where there are more than one constraint surfaces cannot be done easily because the estimation of the constraint's orientations is much harder in this case. Furthermore, if the constraints change abruptly, unwanted transients will be created because of the control mode switching. The use of reasoning with approximate or partial models of the environment appears to offer a solution to these problems.