New robot learning method combines RL with control theory

Researchers at the University of Illinois Urbana-Champaign combined reinforcement learning with control theory to guarantee robot stability. The method improves tracking performance by up to 15% over existing approaches and was tested on wheeled, flying and six-legged robots.

• Method pairs reinforcement learning with control theory to guarantee stability
• Tracking performance improves by up to 15% over existing methods
• Approach tested on wheeled, flying and six-legged robots