Smart bicycle knows when to help and when to back off

Pic/Shibaura Institute of Technology

Researchers at Japan's Shibaura Institute of Technology have developed a steer-by-wire bicycle that can tell whether a rider is turning or wobbling.

The system uses machine learning to read steering angle, speed, roll, lateral acceleration, and rider reaction torque. Data from these signals trains a special AI to classify straight riding, cornering, and instability.

While other assistance systems, which may overreact when a bicycle leans into a normal turn, the new setup keeps its stabilisation control switched off during manoeuvres.

Published in IEEE/ASME Transactions on Mechatronics, the research could make e-bikes, electric motorcycles, shared bikes and delivery vehicles safer, especially for older or inexperienced riders.