Description

This project presents an alternative algorithm for real- time path planning with both static and dynamic uncertain obstacles in an environment. We used a real-time sample-based method called Rapidly Exploring Random Tree algorithm (RRT*) combined with the Potential-based algorithm for dynamic obstacle avoidance. To be more specific in potential-based planner, our algorithm only executes potential-based planner whenever our agent is inside the influence circle of the moving obstacle. The influence circle will change its size depending on the relative velocity and maximum acceleration of the agent. Our version of RRT* starts by expanding the tree from the goal position. This can ensure the path from the agent to the goal point is the most optimal if the agent is intruded by the moving obstacle and derailed from the original path.

Technical Details

RRT* with only static obstacle

This is the simulation of the omnidirectional robot (yellow circle) navigating from the starting point to the endpoint (red circle) while avoiding collision with the static obstacles (blue and green circles) using RRT* as the planning algorithm.

RRT* and potential field algorithm with moving obstacles

This is the simulation of the omnidirectional robot (yellow circle) navigating from the starting point to the endpoint (red circle) while avoiding collision with the static obstacles (blue circles) and the moving obstacles (green circle). The moving obstacle is controlled by a user and it has an influence circle around it (fainted green circle) representing the potential field. The robot will navigate to the endpoint using RRT* as the planning algorithm and potential field for obstacle avoidance. You can see the algorithm switches (bottom left text) while the robot is within the influence circle of the moving obstacle.

If you would like to know more about the project, please contact me through: yychan@bu.edu

Real-Time Sample-based Planner with Artificial Potential Field for Dynamic Obstacle Avoidance

Description