Interaction and Decision-Making in Autonomous-Driving

A (Virtual) Workshop at RSS 2020

Workshop Description

As an ever-growing number of autonomous vehicles are deployed on public roads, developing robust decision-making algorithms becomes increasingly important. To enable better decision making, recent research efforts have focused on detecting, modelling, and predicting the behavior of traffic participants. The first aim of this workshop is to discuss recent advances, open research challenges, and future directions for robust decision making for autonomous driving. This includes the requirements and advances for situation-awareness pipelines to enable robust decision making, the scientific challenges involved in integration of these pipelines, and the decision-making aspects involved towards enabling robust interactive autonomy. The second aim of this workshop, is to discuss the interplay between prediction and planning, given the multi-agent nature of driving. "Socially-aware" motion planning w.r.t. forecasting models of other agents is often often necessary, but also vice-versa: forecasting w.r.t. plans, to anticipate how a robot’s plan will likely affect surrounding drivers. Given the increasing amount of interest in this area in robotics, computer vision, and machine learning communities, we hope this workshop can be a suitable venue to promote further discussion and developments in this area.

COVID-19 and a Virtual conference: RSS will be a virtual conference this year, and we are working with RSS organizers to organize teleconferencing options. We will relay all updates here. Any questions please email: rmcallister@berkeley.edu

Call for Contributions

Interested researchers from both academia and industry are invited to submit papers in the following format: 4 pages + references + supplementary (if needed), as a single PDF. Successful papers will be presented in a poster session, and possibly spotlight talks. We welcome any submissions, including those submitted to the RSS conference itself.

Submission Website: https://cmt3.research.microsoft.com/IDA2020
LaTeX Template: https://roboticsconference.org/docs/paper-template-latex.tar.gz

Since RSS recently announced it will be transitioning to a virtual conference, we've postponed the submission deadline to allow authors more time:

Dates

  • Paper due: 18th June 2020 AoE

  • Notification: 2nd July 2020

  • Workshop: 13th July 2020

The topics of interest involve, but are not limited to:

  • Multi-agent reinforcement learning

  • Motion forecasting of interactive behavior

  • Intent prediction

  • Socially-aware planning

  • Motion planning under uncertainty

  • Inverse reinforcement learning

  • Causal modelling for multi-agent systems

  • Game theory


  • POMDP planning

  • V2V communication and coordination

  • Gesture recognition

  • Forecasting metrics and benchmarks

  • Human-robot collaboration

  • Situation Awareness

  • Dynamic Scene Understanding

Confirmed Speakers

Toyota Research Institute

Stanford

Tsinghua University