2nd Workshop on Informative Path Planning and Adaptive Sampling

Organizers: Jen Jen Chung, Nicholas Lawrance, Graeme Best, Alberto Quattrini Li, Stephanie Kemna

Website: https://n.ethz.ch/~chungj/WIPPAS2019/

Data-derived models form a core component of robotics, both as input, i.e. internal representations of robots and the environment, and output, i.e. scientific data gathering. Robots offer unique capabilities for collecting data that was previously too time consuming, dangerous or infeasible for humans, and on a larger scale than was previously possible. However, not all data is created equal, and how well a data-derived model represents the true underlying phenomenon is directly dependent on the quality of the samples. In applications with constraints on the amount of data that can be collected, it is therefore important to maximize the quality of collected data. This is the underlying problem in informative path planning and adaptive sampling. Despite the prevalence of adaptive sampling methods in state-of-the-art applications, e.g. environmental modeling or reinforcement learning, there are still many open problems. For example: How can we best handle modeling inaccuracies, incorporate expert knowledge, or design multi-robot information sharing protocols to reason over joint sampling spaces?

The theme of informative path planning and adaptive sampling spans all robotic domains and we aim to assemble researchers from many fields—e.g. marine, ground, and aerial robotics, multi-agent, and learning communities—to improve cross-domain awareness. We will look at various aspects of informative path planning, including its theoretical foundations, active sampling, multi-robot planning, and its application to real-world problems. This workshop will build on its successful predecessor WIPPAS’18, held at ICRA 2018.