Excitable Integral Reinforcement Learning (EIRL)

This algorithm is developed in the field of control systems for control of affine nonlinear continuous time (CT) systems, which include important application areas such as robotics and autonomous systems, aerospace, manufacturing, and process control. This invention falls under the class of continuous-time reinforcement learning (CT-RL) control algorithms. CT-RL algorithms hold great promise in real-world control applications. Adaptive dynamic programming (ADP)-based CT-RL algorithms, especially their theoretical developments, have achieved great successes. However, these methods have not been demonstrated for solving realistic or meaningful learning control problems. This work leverages various ideas from the well-established field of classical control, in particular Kleinman's algorithm, plus learning from nonlinear state/action data to achieve robust and efficient control design with improved time/data efficiency and well-behaved system responses.

 

Researchers at Arizona State University have developed an Excitable Integral Reinforcement Learning (EIRL) platform. Excitable Integral Reinforcement Learning is an advanced set of algorithms designed to control continuous-time affine nonlinear systems by improving persistence of excitation (or learning exploration) and decentralizing control problems to reduce computational complexity. It addresses the numerical instability issues of existing ADP-based reinforcement learning methods, enabling robust and efficient control of complex, realistic systems such as hypersonic vehicles. EIRL guarantees convergence to optimal control laws and closed-loop stability, making it ideal for mission-critical applications.

 

Potential Applications

Benefits and Advantages

For more information about this opportunity, please see