As a project in my graduate motion planning class, my team developed a potential field algorithm to generate random paths through a room for trackless dark ride vehicles. We chose the potential field method because it is able to create smoother paths.
Two example plans are shown here. The green circle indicates the vehicle, and the white arrow inside indicates the orientation, which is not necessarily the direction of travel. The purple shows areas where the center of the vehicle can go without colliding with walls. The red shows areas that would cause collision with obstacles.
To generate the paths, we selected several random points, which are shown with stars. These points were chosen such that they were not all nearest to the same obstacle, ensuring that the vehicle will move around the entire room. These points are represented by stars in the plan. The potential fields sequentially planned the route from one of these points to the next. The car’s orientation was chosen so that it was directed towards the obstacle that was nearest to the goal it was approaching. The algorithm is able to generate a new plan about every 10 seconds. The detailed report for this project can be found here.
This work included the following:
1) Potential field motion planning
2) Algorithm modification
3) Path smoothing