Explainable and Trustworthy Robot Decision Making for Scientific Data Collection

An RSS 2020 Workshop


Robotic information gathering has many applications in fundamental research (e.g. oceanography, geology, biology) and in resource extraction (e.g. mining, agriculture, forestry). The state of practice is for humans to oversee robotic operations, often with monitoring burdens that can have significant costs. Remote operations are plagued by restrictions on communications: bandwidth limits, latency, and drop out. There is a need to develop systems for scientific data collection that operate autonomously with high neglect tolerance.

Building trustworthy autonomous systems hinges on a multitude of complex factors, such as humans’ ability to communicate their actual objectives to robotic systems, robots’ ability to explain their behaviour to remote operators, and users’ previous experience with deployed robots. Automated science operations demand more than just specialist training with robots. They demand a fundamental understanding of (and new technologies to address) difficult problems, such as Explainable AI and Competency Aware reasoning, given the non-obvious requirements for correct behaviour.

This workshop will bring together roboticists and scientists to identify what roboticists need to understand about science objectives, as well as what capabilities roboticists can provide to achieve those objectives. We are looking for submissions that address: the problem of trusting autonomous data collection for short- and long-term operations, methods to elicit scientists’ preferences, and gaps between science needs and autonomous capabilities. The outcome of this workshop will be organized into a report documenting the major conclusions of the workshop and a potential special issue in a major robotics journal (if selected by the journal editorial boards).

Workshop Details

Call for Submissions

We are looking for 2-page abstracts in PDF form. Selected submissions will be presented as posters during the interactive sessions of the workshop.

Abstract Submission Deadline: 9 April 2020 Anywhere on Earth

Notification for Contributors: 16 April 2020

Submission Process: Submission via Easy-Chair

Date and Location

Date: 13 July, 2020

Time: 0800hrs to 1800hrs

Location: Corvallis, Oregon


E-Mail: explainable.data.collection@gmail.com

Topics of interest for this workshop:

Trust in Autonomous Systems

  • Explainable AI
  • Machine Self-Confidence
  • Formal Validation and Verification

Autonomous Data Gathering Applications

  • Environmental monitoring
  • Planetary exploration
  • Search and rescue
  • Disaster recovery
  • Expert preference elicitation
  • Interpretable models of decision making
  • Hypothesis-based reasoning and hypothesis generation



Organizing Committee

Nisar Ahmed, University Colorado, Boulder

P. Michael Furlong, NASA Ames Intelligent Robotics Group [KBR, Inc.]

Geoff Hollinger, Oregon State University

Seth McCammon, Oregon State University