This paper addresses a general class of problems for sensor-based path planning and exploration for robots moving in unknown environments. The robot is assumed to be equipped with a large number of proximity sensors mounted on its body (a "skin" sensor). Robot's own motion is used to sense the free space, analogous to a blind person "groping and exploring" using his sense of touch. This sensed free space can be memorized and used to further plan the robot's motion. Using this key idea, we propose a general framework for such skin sensor based motion planning that is valid for robots with large degrees of freedom. The specific motion-planning algorithm developed uses a variant of the probabilistic roadmap method. A novel "metric" based on the notion of C-zone map for robot's movement that results in efficient exploration of configuration space is proposed and implemented. Planar simulations demonstrate our results.