Learning from demonstration provides an attractive solution to the problem of teaching autonomous systems how to perform complex tasks. Reward learning from demonstration is a promising method of inferring a rich and transferable representation of the demonstrator's intents, but current algorithms suffer from intractability and inefficiency in large domains due to the assumption that the demonstrator is maximizing a single reward function throughout the whole task. This paper takes a different perspective by assuming that the reward function behind an unsegmented demonstration is actually composed of several distinct subtasks chained together. Leveraging this assumption, a Bayesian nonparametric reward-learning framework is presented that infers multiple subgoals and reward functions within a single unsegmented demonstration. The new framework is developed for discrete state spaces and also general continuous demonstration domains using Gaussian process reward representations. The algorithm is shown to have both performance and computational advantages over existing inverse reinforcement learning methods. Experimental results are given in both cases, demonstrating the ability to learn challenging maneuvers from demonstration on a quadrotor and a remote-controlled car.

Reward learning from demonstration is the task of inferring the intents or goals of an agent demonstrating a task. Inverse reinforcement learning methods utilize the Markov decision process (MDP) framework to learn rewards, but typically scale poorly since they rely on the calculation of optimal value functions. Several key modifications are made to a previously developed Bayesian nonparametric inverse reinforcement learning algorithm that avoid calculation of an optimal value function and no longer require discretization of the state or action spaces. Experimental results given demonstrate the ability of the resulting algorithm to scale to larger problems and learn in domains with continuous demonstrations.

Inverse reinforcement learning (IRL) is the task of learning the reward function of a Markov Decision Process (MDP) given knowledge of the transition function and a set of expert demonstrations. While many IRL algorithms exist, Bayesian IRL [1] provides a general and principled method of reward learning by casting the problem in the Bayesian inference framework. However, the algorithm as originally presented suffers from several inefficiencies that prohibit its use for even moderate problem sizes. This paper proposes modifications to the original Bayesian IRL algorithm to improve its efficiency and tractability in situations where the state space is large and the expert demonstrations span only a small portion of it. The key insight is that the inference task should be focused on states that are similar to those encountered by the expert, as opposed to making the naive assumption that the expert demonstrations contain enough information to accurately infer the reward function over the entire state space. A modified algorithm is presented and experimental results show substantially faster convergence while maintaining the solution quality of the original method.