Optimizing dynamic locomotion in Baleka II: from simulation to real-world running

Martin, Zubair

Optimizing dynamic locomotion in Baleka II: from simulation to real-world running

Thesis / Dissertation

2025

Publisher

University of Cape Town

Department

Department of Electrical Engineering

Faculty

Faculty of Engineering and the Built Environment

Abstract

In the field of legged locomotion, agility is a critical area of research in robotics due to its potential to enable versatile movement for various applications, including search and rescue missions. However, bipedal robots face significant challenges in achieving rapid movements, such as maintaining stability and agility. This dissertation presents the development of Baleka II, a bipedal robot designed to overcome these challenges by achieving rapid legged locomotion through open-loop control. Building upon its predecessor, this research seeks to evaluate the robot's capacity to perform agile tasks by incorporating trajectory optimization algorithms and conducting real-world experiments. The study is structured around four primary objectives: improving the embedded system configuration, generating control trajectories using trajectory optimization, validating these solutions through simulations, and implementing them on the physical robot. The key locomotive tasks investigated include acceleration, deceleration (gait termination), and steady-state walking/running. The control system was implemented using the Speedgoat Real-Time Target Machine, integrating Simulink Real-Time and Simscape Multibody for real-time execution. Trajectory optimization was accomplished using Pyomo and Interior Point Optimizer (IPOPT), producing solutions for walking (0.5 m/s), walk-to-run transitions (1.5 m/s), and maximum forward speeds (4.0 m/s). Simulations were used to verify these solutions, taking into account the robot's physical constraints. Despite the use of open-loop control, stability was maintained through proportional-derivative (PD) controllers for each motor. The key findings of this research indicate that as the robot's speed increased, so did the actuation effort, peak torque, and GRFs, leading to velocity discrepancies and high deceleration upon ground contact. Nevertheless, Baleka II was able to accelerate into 3.2 m/s steady-state gait and decelerate in a stable manner, demonstrating competitive acceleration and deceleration rates relative to other bipedal robots. These results offer valuable insights into the use of open-loop optimal control for achieving rapid transitions in bipedal robots, with potential applications in search and rescue, industrial assistance, and entertainment. Future work will focus on enhancing the robot's deceleration capabilities, integrating additional sensors, exploring advanced control techniques, and testing the robot on uneven terrain. These efforts will further expand the potential of Baleka II for real-world applications.

Keywords

engineering

Reference:

Collections

Masters

Full item page