Dynamic motion planning for autonomous and semi-autonomous mobile robot navigation

Course : Directed Research

Problem Statement : To develop a novel dynamic obstacle avoidance approach for assistive-autonomy of autonomous and tele-operated mobile robots with dynamic collision and obstacle avoidance

Application : Tele-operated nursing robots to serve quarantined patients in hosppitals

Approach :

• Developed a dynamic motion planning approach based on planning in velocity spaces instead of conventional path planning.

• Employed the use of Optimal Reciprocal Collision Avoidance (ORCA) for velocity planning. The algorithm, rather than giving a set of points for navigation, gives instantaneous robot velocity close to the desired velocity which if followed by the robot avoids collisions with all obstacles around it.

• Generated static environment info by employing SLAM and image processing to generate obstacle and edge data.

• Using Kalman filter based approach for object detection and motion estimation with 2d LiDAR sensors generated dynamic obstacle data

• Achieved robust and complete static obstacle avoidance in different environments and currently working to achieve dynamic obstacle avoidance