Opportunities and Challenges from Using Animal Videos in Reinforcement Learning for Navigation
This work addresses efficiency in RL navigation for robotics or AI systems, but it appears incremental as it builds on existing methods with a new data source.
The paper tackled improving reinforcement learning for navigation with sparse rewards by using animal videos, showing that their methods enhanced performance over algorithms without such observations.
We investigate the use of animal videos (observations) to improve Reinforcement Learning (RL) efficiency and performance in navigation tasks with sparse rewards. Motivated by theoretical considerations, we make use of weighted policy optimization for off-policy RL and describe the main challenges when learning from animal videos. We propose solutions and test our ideas on a series of 2D navigation tasks. We show how our methods can leverage animal videos to improve performance over RL algorithms that do not leverage such observations.