An integrated motion planning approach for virtual human arm manipulation

Planning manipulation path for virtual human arm takes an important role in such areas as computer animation and virtual maintenance engineering. The problem is usually defined to find a feasible non-collision path to connect initial configuration and goal configuration in joint configuration space. Since users are more concerned about the position and orientation of end-effector for reaching and grasping tasks, inverse kinematics (IK) is often adopted to calculate a non-collision goal configuration for planning procedure. Though methods based on geometrical calculation and non-linear programming have been proposed to solve IK positioning of rigid segments of virtual human arms, the solution might result in infeasible or unreachable goals located in regions that are disconnected from initial configuration. In this paper, a novel integrated motion planning approach that combines the workspace goal-orientated heuristics and a variant of Rapidly-exploring random tree (RRT) algorithm is presented to efficiently planning motions for virtual human arms without the need of IK solutions. Experimental results demonstrate the effectiveness of the approach.