Intention-aware cooperative driving behaviour model for automated vehicles

This innovative ARC Discovery project, in collaboration with Partner Investigator Wendy Ju at Cornell University, expects to generate knowledge about a new Cooperative Driving Behaviour model for Automated Vehicles (AVs), utilising a transdisciplinary approach that mixes human-centric methods with cutting-edge deep learning techniques. It uses Intention Awareness (IA) as an innovative investigative lens to explore driving as a cooperative task.

The ideal cooperation between driver and Autonomous Vehicle (AV) has often been illustrated with the H-metaphor, where “H” stands for “horse”. The idea is that drivers should be able to collaborate and communicate with the AV in a similar way to how riders communicate with their horse. We argue that the H-metaphor is not the best for future AVs and propose a more human-centric one, analogous to what PI Wendy Ju refers to as the Husband-metaphor in her book The Design of Implicit Interactions: “My husband is an autonomous driver. I trust him to drive […], I never worry once about his capabilities. However, when I sit next to him in the passenger seat, I also participate in driving. I help make decisions about where to go, and suggest alternative routes to take. I warn about potential issues and point out latent hazards that I think my husband might not see.” In addition, passengers may also be receptive to cues that tell us how aware the drivers are of the current driving situation, e.g., the husband taking the foot off the pedal and hovering over the brake pedal, which indicates his intention to (potentially) brake, and by doing so conveying his awareness (or lack thereof) that a pedestrian may cross the street.

This project will therefore:

  1. Define a new computational model for Cooperative Driving Behaviour for automated vehicles based on joint-intention awareness concepts
  2. Create new Human Machine Interfaces (HMI) designed to share intention and facilitate cooperation
  3. Explore innovative methodologies to advance future cooperative behaviour research

Intended outcomes are

  • new joint intention awareness theory,
  • new interfaces for automated vehicles,
  • new methodology for cooperative behaviour research, and
  • enhanced research capacity.

The expected significant benefits are for automated systems to become more predictable, acceptable, readable and safer to use by everyday people.

Media

QUT-led project to fast-track deployment of automated vehicles


Funding / Grants

  • ARC Discovery Grant (2018 - 2020)

Team

Other Team Members

This project is a collaboration with Partner Investigator Wendy Ju at Cornell Tech in New York.

Partners