Control-aware trajectory planning slashes robot actuator effort in new study

A new paper from researchers Bhanuka Dayawansa and Rohan Munasinghe challenges the assumption that kinematically smooth trajectories are efficient for robot manipulators. Classical planners like cubic and quintic splines generate smooth paths but ignore the robot's dynamics and control effort, leading to inefficient nonlinear execution and excessive corrective torque. The authors propose a control-aware optimal trajectory planning framework that explicitly incorporates the manipulator's nonlinear dynamics and actuator constraints within a finite-horizon optimization. A key innovation is a midpoint linearization strategy that improves approximation accuracy for large point-to-point motions, enabling fair comparison under identical closed-loop nonlinear execution conditions.

Simulations on a simplified UR5 manipulator show the new approach consistently outperforms classical methods across all scenarios. It achieves substantial reductions in tracking error, corrective torque, and overall closed-loop execution cost. The unified evaluation framework ensures that differences come solely from trajectory generation, not from controller tuning or dynamics modeling. The findings underscore that kinematic smoothness does not guarantee dynamically efficient execution, and that control-aware planning is essential for practical robotics applications. The paper has been accepted for presentation at MERCon 2026.