A continuous RRT*-based path planning method for non-holonomic mobile robots using B-spline curves

Abstract

Rapidly exploring random trees (RRT) are sampling-based approaches being widely applied for path planning of mobile robots. Since the output of these algorithms usually is a stream of discrete lines involving discontinuity at the linking points, kinematic constraints restrict the robot's movements. Consequently, robots may not pass discrete points in the path correctly. Hence, the using CAGD (Computer-Aided Geometry Design) curves can run simultaneously alongside those algorithms or may run after that to make a smooth path and that's the way in which non-holonomic constraints can be considered perfect and robots can be droved autonomously across them about the collision detection method which executed by the main sampling-based algorithm like RRT*. In this paper, an approach based on the combination of RRT* and B-spline is proposed for smoothing the path which is generated by RRT*-based algorithms, which are one of the most famous groups of algorithms in artificial intelligence. Some new functions are added to the outcome of the RRT* algorithm. To avoid collision in the generated path, some corrections are also provided. Finally, for illustrating the efficiency of the proposed method, the algorithm is implemented in the simulation environment of Webots^® and for verification, the obtained results are compared and discussed.