The paper deals with the problem of motion\udplanning of anthropomorphic mechanical hands avoiding collisions.\udThe proposed approach tries to mimic the real human\udhand motions, but reducing the dimension of the search space\udin order to obtain results as a compromise between motion\udoptimality and planning complexity (time) by means of the\udconcept of principal motion directions. Basically, the work\udincludes the following phases: capturing the human hand\udworkspace using a sensorized glove and mapping it to the\udmechanical hand workspace, reducing the space dimension by\udlooking for the most relevant principal motion directions, and\udplanning the hand movements using a sampling-based roadmap\udplanner. The approach has been implemented for a four finger\udanthropomorphic mechanical hand, and some examples are\udincluded to illustrate its validity.
展开▼
