Robots for Understanding Natural Ecosystems
[RUNE Workshop]

May 17th, 2024

at ICRA 2024 Yokohama, Japan

ABSTRACT

Natural ecosystems are defined by complex webs of interactions between the flora, fauna, and non-organic elements of an environment. The importance of modeling and understanding these entangled and dynamic interactions is increasing as anthroprogenic change threatens ecosystems across the globe. Marine, terrestrial, and aerial robotic platforms increasingly capable of safely navigating remote environments over extended periods provide opportunities to better and quickly study such ecosystems. These platforms leverage advances in field robotics, computer vision, and artificial intelligence, such as semantic mapping, reinforcement learning, and distributed or adaptive planning. With these, robots can be more than just mobile sensors collecting data for human analysis; they can actively and independently aid in exploration and understanding across a diverse range of ecosystems and synthesize complex high-level concepts into targeted and actionable plans for conservation and science.

This workshop brings together robotics researchers at the forefront of environmental sensing and exploration, animal monitoring, semantic modeling, and more to present a panoramic view of robotic advances in the pursuit of understanding natural ecosystems. We further invite researchers in ecology and related fields, who use or wish to use robotics, to attend and contribute their insights on the most critical challenges and opportunities for greater use of autonomous robots to solve these problems.

CRITICAL QUESTIONS

What are the scientific gaps of knowledge that need to be addressed concerning ecological understanding? At what spatiotemporal scales and across which domains should roboticists be targeting their innovations?
Ecological data is often high-dimensional and deeply entangled, how do we synthesize these into actionable quantities to better inform short-term robot and long-term conservation decision-making and planning?
What metrics are used for quantifying ecological health or success that can be leveraged by roboticists and vice versa? Similarly, are there simulators, datasets or benchmarks that can assist in these developments and to what extent can they be trusted (or not)?
How can the latest algorithmic strategies in multi-robot coordination, semantic mapping, distributed planning, and machine learning be used to address these gaps and how can we deploy them in the field?
How can we bring robotics research and ecological research more in line with each other and how can we form lasting collaborations?
What funding sources are available to support roboticists and ecologists to collaboratively develop scalable ecosystem monitoring solutions?