The Mini Wheelbot: A Testbed for Learning-based Balancing, Flips, and Articulated Driving

TL;DR

The paper presents the Mini Wheelbot, a compact reaction-wheel unicycle designed as a robust testbed for learning-based control of unstable, nonlinear dynamics with automatic environment resets. It couples a nonlinear yaw-driving framework with a balanced state-feedback controller and a nonlinear MPC, demonstrated through Bayesian optimization for controller tuning and imitation learning via an approximate MPC that enables onboard articulated driving. Key contributions include a full hardware/software redesign, a parameter-identification workflow, and two learning-based experiments that showcase automatic reset capabilities and fast onboard control. The work advances practical learning-driven control in robotics by providing a versatile platform for testing, benchmarking, and extending learning methods in a challenging, real-world setting.

Abstract

The Mini Wheelbot is a balancing, reaction wheel unicycle robot designed as a testbed for learning-based control. It is an unstable system with highly nonlinear yaw dynamics, non-holonomic driving, and discrete contact switches in a small, powerful, and rugged form factor. The Mini Wheelbot can use its wheels to stand up from any initial orientation - enabling automatic environment resets in repetitive experiments and even challenging half flips. We illustrate the effectiveness of the Mini Wheelbot as a testbed by implementing two popular learning-based control algorithms. First, we showcase Bayesian optimization for tuning the balancing controller. Second, we use imitation learning from an expert nonlinear MPC that uses gyroscopic effects to reorient the robot and can track higher-level velocity and orientation commands. The latter allows the robot to drive around based on user commands - for the first time in this class of robots. The Mini Wheelbot is not only compelling for testing learning-based control algorithms, but it is also just fun to work with, as demonstrated in the video of our experiments.

Figures (9)

• Figure 1: The Mini Wheelbot: A small, rugged, and symmetric reaction wheel unicycle robot with challenging nonlinear, unstable, and hybrid dynamics. The Mini Wheelbot can stand up from any position which allows for automatic environment resets in learning-based control experiments.
• Figure 2: System overview of the Mini Wheelbot.
• Figure 3: Electronics design of the Mini Wheelbot (top) and custom circuit boards (bottom).
• Figure 4: Control system overview and generalized coordinates.
• Figure 5: Five of the trajectories used for parameter identification: measurements (dotted) of yaw angle (top) and yaw rotational velocity (bottom) and predictions with optimal parameters $p^*$ (solid).