Lane Change Path Planning at High Speeds using Neural Network

Lane change and braking maneuver are two life savers at high speeds. In this paper, first the required speed for lane change is obtained, using the braking maneuver. Then lane change path planning is provided in three steps: 1) specifying the allowable path planning space, 2) generating the possible path(s), 3) selecting the final path. For this purpose, three separate neural networks have been used which are trained based on an exact model of vehicle, retrieved from CARSIM. This papers important innovation is the path planning capability in a wide range of road friction factors and widths. Also low computation cost is a benefit of using neural networks, which is vital at high speeds. The results show that the provided path planning algorithm has an almost 9% error.