A novel virtual force field method based on obstacle avoidance via polynomial interpolation

The key point of a autonomous position changing for a vehicle between moving or stationary obstacles is that being able to reach the target point without any collisions with the obstacles. In this thesis work, a new method for solving the obstacle avoidance problem, which is fundamental to autonomous movement, is presented. In this method, the future motion behavior is predicted by polynomial extrapolation to the vehicle's recent path history, and it is determined whether a collision with obstacles in the environment will occur. This presented method is integrated with the virtual force field theory, which is studied by many academic studies in obstacle avoidance, and a simulation environment is established. First, a simplified dynamics of an autonomous vehicle with stationary obstacles is considered. To ensure the vehicle avoids collisions under combined attractive and repulsive forces, we propose a polynomial extrapolation using its past trajectory to predict its future trajectory. As the extrapolated trajectory intersects with static obstacles in the environment, we include these obstacles in the virtual force field calculations, considering both time and trajectories to guide the vehicle in avoiding obstacles. Secondly, simplified dynamics of an autonomous vehicle with dynamic obstacles is considered. Obstacle avoiding maneuver of the vehicle is observed if the extrapolated trajectory of dynamic obstacle intersect with the vehicle's extrapolated trajectory.