Several designs for a spherical rolling robot have been suggested and some of them were implemented. The kinematics and dynamics study as well as the path planning for the rolling robot are based on the assumption that the deterministic model describes the actual rolling robot successfully. However, due to the high mobility of the sphere, the stochastic behavior is obviously observed. In this paper, we first build a rolling robot to confirm the stochasticity. The robot is actuated by a mass-shifting mechanism where an unbalanced weight inside the spherical robot is rotated by two motors, and the imbalance induced by the weight makes the robot roll. After confirming that this actual rolling robot shows the stochastic behavior, we propose a path planning method for the spherical robot rolling on the plane. The path-of-probability method is applied to generate the most probable path from starting location to destination. This planning method uses the stochasticity of the system to produce the probability density function, and generates the piece-wise short steps for the robot move, which construct the whole trajectory that the robot should follow.
