Reproduce “Target Search and Navigation in Heterogeneous Robot Systems with Deep Reinforcement Learning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yun Chen, Jiaping Xiao
A deep reinforcement learning system enables collaborative search and rescue operations for a UAV and a UGV in simulated underground mine environments, where the UAV identifies targets and the UGV navigates to them by following the UAV's path. The system achieves an 89.1% success rate in the training environment and 67.6% in a more complex generalization environment.
Key insights
- Decomposing the learning problem into a multi-stage training framework significantly improves training stability and convergence speed for collaborative robot systems.
- Utilizing an Intrinsic Curiosity Module (ICM) effectively addresses the sparse reward problem by providing intrinsic motivation for exploration, leading to robust learning.
- Heterogeneous robot collaboration allows leveraging complementary strengths, with the UAV's superior sensing for search and the UGV's stability for navigation to the target.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.