A new approach to controlling an active suspension system based on reinforcement learning

Active suspension provides better vehicle control and safety on the road with optimal driving comfort compared to passive suspension. Achieving this requires a good system that can adapt any environment. This article uses deep reinforcement learning method develop an neural network meets requirements according ISO 2631-5 standards. The algorithm trains agent without prior knowledge of Various simulations were performed, results validated literature standard until appropriate reward function was found. Simple consistent profiles used while maintaining constant parameters during training. show based reduces body acceleration improves ride sacrificing deflection dynamic tire loading. controller expects RMS value be 0.228 minimum overrun suspended mass.