RESEARCH ON REACTIONLESS MANIPULATOR

点击翻译按钮获取中文摘要

We developed a general method for reaction compensation via inertial coupling. This method we called "reaction null space" method. In our early work we focused on moving base systems, such as free-flying space robots and long-reach manipulators or flexible base manipulators. We have shown that via the reaction null space, the dynamics of the manipulator are decoupled from the dynamics of the base. Using the notation, we can analyze the system, for example, to obtain the so-called "reactionless paths" in the workspace of the manipulator. Any motion along these paths will not excite the base dynamics.Also, the reaction null space is useful for design purposes. This is clear from the fact that the reaction null space may exist with different types of design. In our latest work we extend the scope of the reaction null space method to cover the whole class of underactuated manipulators. In addition to moving base robots, this class includes also free-and flexible-joint robots and flexible-l … More ink robots.We designed a planar experimental system of a flexible-base manipulator comprising a flexible base with an elastic cantilever beam. Initially, a small 2R rigid link manipulator was mounted on the tip of the base. A force/torque sensor with a special design was mounted on the base of the rigid link manipulator to measure reactions in the plane. Control feedback from elastic base deflection was obtained through a strain gauge. With this design we could obtain useful experimental data for the one-arm case, which confirmed our expectation regarding reactionless motion. Later, we attached a second rigid-link manipulator to the system. With this experimental setup we could perform noncontact motion experiments, as follows.First, feasibility of reactionless path tracking was confirmed both under open loop and closed-loop control.In either case, the disturbance of the base was ignorable. This situation didn't change when the acceleration and the speed along the reactionless path were changed. This is well in agreement with theory. We note that the reactionless paths do exist also for a dual-arm manipulator holding an object. This was verified experimentally. Second, we developed a three-phase path planning method for a single arm. The experimental performance confirmed our expectations. We compared the results with the well-known method of Disturbance Map. Finally, we implemented vibration suppression control by active damping. We have shown that vibration suppression is decoupled from the reactionless path motion. This means that feedback gains of the two control subtasks can be designed independently. It is possible to suppress vibrations while moving along a reactionless path. We have also confirmed the possibility of vibration suppression with a dual-arm manipulator holding an object. Less