Abstract: Resilient intelligent systems introspect, adapt, and evolve to changing robot and environment models and application objectives. A fundamental challenge to achieving resilient operation is the development of robust techniques that enable online learning and adaptation while preserving the performance guarantees required to ensure safe and stable autonomy.
The goal of this workshop is to discuss the fundamental challenges to achieving resilience in intelligent robots given online learning and adaptation within the feedback loop. The workshop brings together expertise in the areas of learning, planning, perception, and control to discuss the implications of recent advancements in online learning and experiential techniques to broader robotics areas that leverage learned models but require additional performance guarantees to ensure safe operation. Researchers from academia and industry working in topical areas such as active perception, human-robot interaction, manipulation, and multi-robot coordination will discuss the fundamental challenges that arise in pursuit of resilient intelligence and robust autonomy.
The workshop will consist of topical presentations and a panel discussion by a cross-disciplinary group of experts with the goal of highlighting the common challenges that arise due to online learning and adaptation (at the levels of planning, perception, and control) within the autonomy feedback loop and the performance implications. A poster session will highlight recent research advancements.
Intended audience: The workshop is expected to be of interest to researchers in academia and industry that develop autonomous systems and employ online learning methodologies to enable resilient operation. As the workshop explores the role of learning within the feedback loop and the need for methodologies that ensure performance guarantees, the workshop is expected to be of interest to the broader RSS community (those working in learning, perception, planning, and control).