CEBO-Based Stair-Climbing Wheel Design

T-pedal stair-climbing wheel design with CEBO and CoppeliaSim.

CoppeliaSim stair-climbing wheel-design evaluation for CEBO
CoppeliaSim stair-climbing wheel evaluation clip for CEBO
CoppeliaSim stair-climbing evaluation.
Physical stair-climbing test clip for the T-pedal wheel robot with manual guidance
Physical stair test with manual guidance.
Physical stair-climbing test clip for the T-pedal wheel robot
Physical stair test.

I defined a six-variable T-pedal wheel design problem for stair-climbing time and body-pitch stability. After validating CEBO on 2D-12D benchmark functions, I automated the optimization loop with CoppeliaSim evaluations.

  • Role: T-pedal wheel search-space design, CEBO optimization experiments, CoppeliaSim evaluation-loop automation
  • Tools: Bayesian optimization, CoppeliaSim, Python, CAD
  • Result: nominal traversal time decreased by 23.31% from 9.405 s to 7.213 s, and body-pitch stability median decreased by 37.84% from 1.102 deg to 0.685 deg under the nominal friction condition.
  • Paper: Clustering-enhanced Bayesian optimization (CEBO): A case study on T-pedal wheel design of a stair-climbing robot. Advanced Engineering Informatics, revised manuscript resubmitted after minor revision.
  • Code: kelvin926/T-Pedal-with-CEBO