We introduce a new reinforcement learning approach combining a planning quasi-metric (PQM) that estimates the number of steps required to go from any state to another, with task-specific "aimers" that compute a target state to reach a given goal. This decomposition allows the sharing across tasks of a task-agnostic model of the quasi-metric that captures the environment's dynamics and can be learned in a dense and unsupervised manner. We achieve multiple-fold training speed-up compared to recently published methods on the standard bit-flip problem and in the MuJoCo robotic arm simulator.
@article{arxiv.2002.03240,
title = {Multi-task Reinforcement Learning with a Planning Quasi-Metric},
author = {Vincent Micheli and Karthigan Sinnathamby and François Fleuret},
journal= {arXiv preprint arXiv:2002.03240},
year = {2020}
}