Fast and Compute-efficient Sampling-based Local Exploration Planning via Distribution Learning
This work addresses the challenge of efficient and reliable exploration planning for robots, particularly in compute-constrained scenarios, though it is incremental as it builds on existing sampling-based methods with learned enhancements.
The paper tackles the problem of compute-intensive and high-variance sampling-based exploration planning in robotics by learning the distribution of informative views and information gain from spatial context. It shows up to 28% improvement in exploration performance over classical methods and offers better performance-compute trade-offs for constrained systems.
Exploration is a fundamental problem in robotics. While sampling-based planners have shown high performance, they are oftentimes compute intensive and can exhibit high variance. To this end, we propose to directly learn the underlying distribution of informative views based on the spatial context in the robot's map. We further explore a variety of methods to also learn the information gain. We show in thorough experimental evaluation that our proposed system improves exploration performance by up to 28% over classical methods, and find that learning the gains in addition to the sampling distribution can provide favorable performance vs. compute trade-offs for compute-constrained systems. We demonstrate in simulation and on a low-cost mobile robot that our system generalizes well to varying environments.