Agility, characterized by swift, precise, and adaptive movement in complex environments, is crucial for robots to exploit their full capabilities. An agile robot can interact safely and efficiently with its environment and unlock new potential across a broad spectrum of tasks, from seamless human collaboration to effective search-and-rescue missions. Recent advancements in learning-based approaches have made significant progress in various agility tasks, particularly those demanding rapid adaptation, long-horizon reasoning, and intricate perception. Despite these advancements, fundamental challenges persist, such as sim-to-real transfer, high-frequency reactive control, and safe policy execution.
Building on the success of our first Learning for Agile Robotics workshop (recording) at CoRL 2022, we are excited to organize the second Learning for Agile Robotics Workshop at CoRL 2023. Our goal is to bring together researchers from diverse backgrounds to share insights on learning-based agile robotics, dive deep into challenges and lessons learned, and collectively shape the future of this fast-growing field.
Our workshop will focus on these guiding topics for discussion:
- What defines robot agility and how can we evaluate it within a robotic system?
- Can we extend agile robot results beyond lab environments?
- What unique benefits and challenges in using learning-based approaches for robot agility?
- How can recent advancements in machine learning (e.g., large language models, foundational models, transformers) be leveraged to improve robot agility?
- How can agility enhance other fields of robotics, such as human-robot interaction, autonomous driving, and industrial and home automation?
