Trajectory Planning and Optimization for Human-Robot Collaboration

Collaborative robotics is reshaping the landscape of automation by enabling safe, efficient, and intuitive interaction between humans and robots in shared workspaces. Unlike traditional industrial robots that operate in isolation, cobots are designed to work alongside humans, requiring advanced capabilities in perception, decision-making, and motion planning to ensure seamless cooperation. Trajectory planning in collaborative robotics involves generating feasible paths that respect kinematic constraints, avoid collisions, and adapt to real-time changes in the workspace. This thesis investigates safety strategies for shared workspaces, emphasizing speed and separation monitoring to prevent collisions while maintaining workflow fluency. Several robot stopping strategies are experimentally compared using a 7-degrees-of-freedom (DOFs) Franka Emika Panda robot. The best-performing approach dynamically scales safety zones to improve collaboration fluency and is successfully validated in an industrial case study, automating the precise placement of small components on car rear lamps, meeting both safety and cycle time requirements. Minimum-jerk trajectory planning for generating smooth and predictable robot movements are also investigated. A mixed time–jerk optimization framework is implemented and tested on the same robotic platform, ensuring reduced acceleration and jerk while maintaining task efficiency. An extended method for redundant manipulators further optimizes both timing and joint configurations, minimizing end-effector jerk. Moreover, this thesis proposes a neural network-based trajectory planner that enables the robot to mimic human arm movements. Trained on a custom dataset, the network selects kinematic configurations that yield human-like motion patterns, improving trust, predictability, and cooperation between humans and robots. Simulations and experimental validations demonstrate the effectiveness of this approach in achieving intuitive, human-like robotic behavior.