Learning Robotic Navigation from Experience: Principles, Methods, and Recent Results
This work addresses the problem of real-world navigation for robotics, offering a novel approach that could enhance adaptability and performance, though it appears incremental as it unifies existing methods.
The paper tackles the limitations of geometric approaches in robotic navigation by proposing a machine learning framework that learns from experience to handle complex physical challenges, resulting in systems that improve with more data and reason beyond geometry.
Navigation is one of the most heavily studied problems in robotics, and is conventionally approached as a geometric mapping and planning problem. However, real-world navigation presents a complex set of physical challenges that defies simple geometric abstractions. Machine learning offers a promising way to go beyond geometry and conventional planning, allowing for navigational systems that make decisions based on actual prior experience. Such systems can reason about traversability in ways that go beyond geometry, accounting for the physical outcomes of their actions and exploiting patterns in real-world environments. They can also improve as more data is collected, potentially providing a powerful network effect. In this article, we present a general toolkit for experiential learning of robotic navigation skills that unifies several recent approaches, describe the underlying design principles, summarize experimental results from several of our recent papers, and discuss open problems and directions for future work.