Marija Popović | Robot Learning and Informative Path Planning | Tartan Planning Series

Series overviews and links can be found on our webpage: https://theairlab.org/tartanplannings...

Abstract: Autonomous robots represent a new frontier in a wide range of monitoring and research scenarios. In many applications, including precision agriculture, inspection, and exploration, robots are rapidly replacing conventional methods of data collection at higher levels of precision, safety, and time- and cost-efficiency. However, a key challenge towards full robot autonomy is planning paths to maximise the information gathered in an unknown environment while satisfying platform budget constraints, e.g. energy, time, or travel distance. This problem is also formally known as informative path planning (IPP). In this talk, I will discuss recent advances in IPP that enable autonomous robotic data gathering, focusing in particular on the role of learning-based methods, and outline challenges for future work.