COMPARISON BETWEEN A* BASED SPLINES AND STATE LATTICE PATH PLANNING FOR AUTONOMOUS VEHICLES

Planning a path from source to destination avoiding collisions with obstacles is a basic requirement for navigation for any autonomous vehicle. Path generated using the algorithms should satisfy the constraints posed by the vehicle for which the path is being generated. Along with this, the path should also be smooth enough to avoid any jerky movements by the vehicle. Many algorithms have been designed to solve this problem. Among these algorithms, most of these come under graph search, sampling, interpolating and numerical optimization techniques. In this thesis, we have chosen two algorithms for comparison on various metrics. The first implementation is a graph based technique, A* algorithm, to find a collision free path from source to destination and using b-splines, an interpolating technique to smooth this obtained path. The second implementation is state lattice planner, which discretizes the whole search space and generates feasible trajectories which in-turn are used by A* algorithm to find a smooth path. The results obtained using these two techniques are compared on various performance metrics such as execution time, optimality, arc length, path cost, ability to find path in narrow spaces and feasibility of the generated path. Based on the observations, the execution time of the state lattice planner is less than A* based splines planner. However, the drawback of this approach is that it does not create a shortest path and that the path cost and arc length are greater than that of A* based splines approach