Work

Bias-Policy Iteration Based Adaptive Dynamic Programming for First-Order Fully Actuated Systems