DOP: Deep Optimistic Planning with Approximate Value Function Evaluation
This addresses the challenge of scaling reinforcement learning to real-world robotics with high uncertainties and large state spaces, though it appears incremental as it builds on existing methods like Monte-Carlo tree search and deep Q-learning.
The paper tackles the problem of efficiently learning robot behaviors in complex, high-dimensional environments by introducing DOP, a deep model-based reinforcement learning algorithm that uses learned Q-functions to guide exploration and planning via Monte-Carlo tree search, achieving good performance and reduced computational demand in tasks like cooperative navigation and robot fetching.
Research on reinforcement learning has demonstrated promising results in manifold applications and domains. Still, efficiently learning effective robot behaviors is very difficult, due to unstructured scenarios, high uncertainties, and large state dimensionality (e.g. multi-agent systems or hyper-redundant robots). To alleviate this problem, we present DOP, a deep model-based reinforcement learning algorithm, which exploits action values to both (1) guide the exploration of the state space and (2) plan effective policies. Specifically, we exploit deep neural networks to learn Q-functions that are used to attack the curse of dimensionality during a Monte-Carlo tree search. Our algorithm, in fact, constructs upper confidence bounds on the learned value function to select actions optimistically. We implement and evaluate DOP on different scenarios: (1) a cooperative navigation problem, (2) a fetching task for a 7-DOF KUKA robot, and (3) a human-robot handover with a humanoid robot (both in simulation and real). The obtained results show the effectiveness of DOP in the chosen applications, where action values drive the exploration and reduce the computational demand of the planning process while achieving good performance.