AI-Powered Robot Bike Learns to Balance and Jump Using Reinforcement Learning

The Robotics and AI (RAI) Institute has developed the Ultra Mobility Vehicle (UMV), a self-balancing robotic bike that can navigate rough terrain and jump onto high surfaces. Reinforcement learning (RL) is enabling the UMV to adapt to complex situations and improve performance.

Credit: RAI Institute

Unlike traditional self-balancing bikes, the UMV does not rely on a gyroscope for stability. Instead, it mimics the movement of a regular bicycle, steering with its front wheel while shifting a weighted top section up and down to maintain balance. RL helps the robot learn movements that conventional control methods struggle to achieve.

One of the biggest challenges for the UMV is riding backward, which is highly unstable. Standard Model Predictive Control (MPC) has difficulty maintaining balance, especially on uneven ground. However, RL allows the robot to adapt and remain stable even in unpredictable conditions.

The UMV is first trained in a simulated environment before being tested in real-world conditions. Simulation enables researchers to refine movement strategies quickly, though transferring these skills to physical robots remains a challenge. By integrating real-world data into simulations, researchers can improve accuracy and reliability.

As RL technology advances, it continues to redefine robotic movement. From enabling a robotic bike to jump onto a table to enhancing humanoid locomotion, RL is expanding the possibilities of robotics and automation.

---

• The RAI Institute has developed the UMV, a self-balancing robotic bike.

• The UMV does not use a gyroscope but balances by shifting a weighted top section.

• RL helps the UMV adapt to complex movements, including riding backward.

Source: CIRCUIT DIGEST