Bicycle Acrobatics with Reinforcement Learning

2026-08-01

AI Summary

This work addresses the challenge of achieving high-agility stunts on bicycle robots, which are inherently difficult due to their simple structure, underactuation, and nonholonomic constraints. The study presents the first application of reinforcement learning to this platform, introducing a multi-task RL framework that integrates waypoint tracking, posture control, payload following, and motion imitation. A state-triggered policy coordinator is designed to seamlessly compose diverse dynamic maneuvers. On the authors’ custom-built UMV bicycle robot, the approach successfully executes complex stunts—including continuous jumps, backflips, wheelies, and three-point turns—enabling it to clear 1-meter-high obstacles, perform over 15 consecutive jumps, and execute action sequences exceeding 20 steps. Both simulation and real-world experiments demonstrate the method’s robustness, generalization capability, and its ability to surpass the agility limitations typical of conventional wheeled platforms.

Abstract

Bicycle robots are fast and energy efficient, but their simple mechanical design and their underactuated and non-holonomic dynamics make highly agile maneuvers difficult to achieve. Here, we use Reinforcement Learning (RL) to enable a bicycle robot to learn and compose a diverse repertoire of dynamic acrobatic stunts. Using different RL formulations such as waypoint following, pose reaching, twist tracking, guided tracking, and motion imitation, the robot acquires autonomous single and multi-table forward and lateral jumps, steerable jumps, front flips, kip-ups, kip-downs, driving, wheelies, bunny hops, and three-point turns. To coordinate these behaviors, we introduce an orchestrator that transitions between policies using state-dependent triggers, enabling robust long-horizon acrobatic stunts. We validate the approach on the Ultra Mobility Vehicle (UMV), a custom bicycle robot, in simulation and hardware. The robot repeatedly traverses tables up to 1 m high, performs more than 15 consecutive autonomous jumps while following waypoints, handles previously unseen multi-table configurations, executes continuous repertoires of kipups, jumps, flips, kip-downs, over more than 20 consecutive trials, and performs more than 10 consecutive autonomous and steerable repertoires of wheelies, lateral jumps, and single-wheel jump downs. These results demonstrate that RL can endow bicycle robots with levels of agility previously associated primarily with legged platforms while preserving the speed and efficiency of wheeled locomotion, establishing a foundation for bicycle acrobatics.

Problem

• bicycle robot
• underactuated dynamics
• non-holonomic constraints
• dynamic acrobatics
• agile maneuvers

Innovation

• Reinforcement Learning
• Bicycle Robot
• Dynamic Acrobatics
• Policy Orchestration
• Underactuated Systems