Inverse Reinforcement Learning With Constraint Recovery
This work addresses the challenge of recovering constraints in inverse reinforcement learning, which is incremental as it extends standard IRL to constrained settings.
The authors tackled the problem of inferring both reward functions and constraints in constrained Markov decision processes from trajectory demonstrations, proposing an inverse reinforcement learning algorithm that reduces the problem to alternating convex optimization and demonstrating its efficacy in a grid world environment.
In this work, we propose a novel inverse reinforcement learning (IRL) algorithm for constrained Markov decision process (CMDP) problems. In standard IRL problems, the inverse learner or agent seeks to recover the reward function of the MDP, given a set of trajectory demonstrations for the optimal policy. In this work, we seek to infer not only the reward functions of the CMDP, but also the constraints. Using the principle of maximum entropy, we show that the IRL with constraint recovery (IRL-CR) problem can be cast as a constrained non-convex optimization problem. We reduce it to an alternating constrained optimization problem whose sub-problems are convex. We use exponentiated gradient descent algorithm to solve it. Finally, we demonstrate the efficacy of our algorithm for the grid world environment.