Study on real-time obstacle avoidance for unmanned ground vehicles

A dynamic path generation approach to find an admissible path in totally unknown surroundings is presented. The proposed approach is a method to find a reasonable trajectory in unknown obstacles. This avoidance algorithm is based on the potential field approach, and there are some differences and innovations from those methods because a navigation function to handle considers not only position of a vehicle, but also velocity and direction of that physical quantity. Although this presented procedure is proceeded on real-time, the fundamental concept of the receding horizon control is used. The finite future states are used to calculate costs, and the states are solved by dynamic equations and constraints of a vehicle. Although this algorithm is uncomplicated, it can find the path efficiently because it uses simple equations to find an admissible trajectory and gives a high reliability for the result. In addition, This paper introduces how to use this algorithm.