Energy-Aware Trajectory Optimization of Industrial Manipulators under Functional Redundancy

June 24, 2026

This dissertation develops and validates a simulation-based methodology for energy-aware trajectory optimization of a UR5-type industrial manipulator. The main contribution is methodological: numerical reductions in an optimization objective can only be interpreted as physical energy savings when the robot model, inverse-kinematics branch, task tolerances, power convention, transcription method, validation policy, and collision geometry are treated consistently. The work first reconstructs a published UR5 energy benchmark and shows that payload, friction, branch selection, and power accounting strongly affect torque and energy estimates. A publication-comparable reconstruction gives E_+ = 219.6 J, close to the reported 220.4 J, while a fully coherent internal convention gives 128.7 J. The optimization framework uses Hermite-Simpson direct collocation with forward- and inverse-dynamics transcriptions, dense-grid process checks, dynamic validation, collision audits, and post-processed positive electrical energy E_+ as the reporting metric. Matched validation shows that inverse dynamics preserves accepted trajectory quality while greatly reducing computation time, with speedups observed on the horizontal benchmark, curved dispensing task, and sparse VF-2 validation case. The validated procedure is applied to three task classes. On the horizontal benchmark, an orientation relaxation with a combined jerk-plus-energy objective reduces E_+ from 214.13 to 161.99 J. On the curved task, energy-aware optimization reduces E_+ from 791.94 to 712.52 J. Collision audits show that feasibility depends strongly on geometry representation. Finally, the sparse VF-2 transfer case confirms that the method extends to semantic key-pose tasks and local base optimization. The results remain local, simulation-based comparisons rather than hardware-validated or globally optimal motions.

Authors in the community:

• 

  J

  João Pedro Pinheiro de Sousa

  ist180977

Supervisors of this institution:

• 

  J

  João Carlos Prata dos Reis

  ist14040

mechanical-engineering - Mechanical engineering

eng - English

Instituto Superior Técnico