Sensor-based path-planning algorithms for a nonholonomic mobile robot

We propose sensor-based path-planning algorithms for a nonholonomic mobile robot. A car-like robot supervised by these algorithms always reaches its target point (position, and orientation) globally while sensing and avoiding uncertain neighbor obstacles locally in a 2-D environment. In general, all the classic sensor-based path-planning algorithms lead a robot to its destination (x,y) quickly in a 2-D unknown search space. Unfortunately, a nonholonomic mobile robot has three degrees-of-freedom (z,y,θ), and consequently its search space has three dimensions. In this paper, we propose two algorithms NH-I·II. NH-I is designed by mixing modified versions of a famous model-based algorithm best-first and the classic sensor-based algorithm Class1. NH-II is designed by mixing modified versions of another famous model-based algorithm depth-first and Class1. BF and DF can investigate around an unknown obstacle whose dimensions are over two. As a result, by the mixture, we obtain three dimensional or more sensor-based path-planning algorithms NH-I·II for a nonholonomic mobile robot. Finally, in order to check a real-time property of both algorithms, we compare their search costs with that of a model-based path-planning algorithm A_NH* for a nonholonomic mobile robot