PREDICTIVE CONTROL OF A ROBOTIC MANIPULATOR IN AN ENVIRONMENT WITH DYNAMIC OBSTACLES

This work falls within the field of robotics and industrial automation, with a focus on advanced manipulator control. The thesis addresses the problem of motion planning in dynamic environments, where moving obstacles may interfere with the robot's task execution. The objective is to develop and validate a Model Predictive Control (MPC) system capable of following a desired trajectory while actively avoiding dynamic obstacles. The proposed strategy integrates kinematic prediction models, a Generalized Proportional Integral Observer (GPIO) for estimating obstacle motion, and safety constraints based on Control Barrier Functions (CBFs). In particular, flexible safety criteria are designed through Control Barrier Function-based Safety Constraints (CBFSCs) with dynamically optimized decay rates, which are included as decision variables within the control problem. This formulation enables real-time adaptation of the safety margins, ensuring a balance between efficiency and robustness. The system is implemented and tested on the Niryo Ned2 robotic arm within the ROS2 and MoveIt2 frameworks. The results demonstrate accurate trajectory tracking and effective management of potential collisions in dynamic scenarios.