Data-efficient Hindsight Off-policy Option Learning
This addresses data efficiency in hierarchical reinforcement learning for robotics, though it appears incremental as it builds on existing option frameworks.
The paper tackles the problem of data-efficient option learning in reinforcement learning by introducing Hindsight Off-policy Options (HO2), which infers option choices from trajectories and trains policies off-policy, outperforming existing methods on benchmarks and showing benefits in 3D robot manipulation tasks from raw pixels.
We introduce Hindsight Off-policy Options (HO2), a data-efficient option learning algorithm. Given any trajectory, HO2 infers likely option choices and backpropagates through the dynamic programming inference procedure to robustly train all policy components off-policy and end-to-end. The approach outperforms existing option learning methods on common benchmarks. To better understand the option framework and disentangle benefits from both temporal and action abstraction, we evaluate ablations with flat policies and mixture policies with comparable optimization. The results highlight the importance of both types of abstraction as well as off-policy training and trust-region constraints, particularly in challenging, simulated 3D robot manipulation tasks from raw pixel inputs. Finally, we intuitively adapt the inference step to investigate the effect of increased temporal abstraction on training with pre-trained options and from scratch.