Robotic collision avoidance in a flexible assembly cell using a dynamic knowledge base

A knowledge-based approach to handle dynamic changes in assembly cell configuration, to identify potential obstacles, and to determine a collision-free path for the robot is discussed. The knowledge base reduces the computation time during the collision detection and path planning stages thereby allowing for real time applications. The use of vision cameras to obtain the current cell configurations is eliminated. Instead, a dynamic knowledge base is designed to store the changes in cell configuration, track movements of the robot, provide a collision-free path, and learn from the previous experience. Given the initial assembly cell configuration, workstation visitation sequence and product data, the current obstacle locations are obtained from the dynamic knowledge base and then the best path for the robot is computed using two static rule bases. The approach is tested using an Intellidex robot operating in an assembly cell. Several experiments have been performed to study the impact of the dynamic knowledge base design and the results are reported