CAN Drive
The CAN Drive project was undertaken by technology firm Seeing Machines with funding support from the ACT Government. The trial aimed to help understand when and why, from both a safety and a regulatory perspective, a driver should be in control rather than the automated vehicle, and to help manage the transition from one to the other with reduced risk. A Tesla vehicle with autopilot functionality was equipped with a Seeing Machines Driver Monitoring System (DMS) and additional sensors to provide a novel dataset on the state and behaviour of the driver when being visually and cognitively distracted in semi-automated driving.
Start and end dates: Start April 2018, results reported September 2019
Lead organisation: Seeing Machines
Location: Canberra, ACT
Additional organisations: ACT Government
Project budget, funding sources, components or breakdown of cost: $1.35m funding support from the ACT Government (to multiple parties)
Contact person: Mike Lenné
Link to project website: CAN drive
Included technologies
- Automated driving and driver support: driver support - continuous (and "partial" automated driving)
Project stages
- Testing for specific performance
Included locations
- Urban motorway
- Private facilities
Benefits directly sought by the project
- Increase public support to assist deployment and adoption
- Commercial return - product / service development
Project scale
- 60 participants (fully licenced ACT drivers, 36 on test-track, 24 on public roads)
- 1 vehicle (Tesla Model S with Autopilot 2.0)
Further details on included technologies
- Seeing Machines Driver Monitoring System (DMS)
- Logged ocular measures at the sampling rate of 46 Hz
- Additional sensors included a time-of-flight camera, two webcams, a forward-facing camera, a Mobileye sensor, a Polar H10 heart rate sensor and an Actigraphy wrist band
Objectives
CAN Drive aimed to deliver insights on when and why, from both a safety and a regulatory perspective, a driver should be in control rather than the automated vehicle, and to help manage the transition from one to the other with reduced risk.
- Drive improvements in vehicle technology and road safety strategy.
- Drive community interest and acceptance of new vehicle technologies.
- Assess the potential for automated vehicle technologies to address social mobility challenges.
- Build Canberra’s reputation as a technology testbed.
Research questions that the project aims to address
- To better understand the impact of different levels of driver distraction on driver behaviour in assisted driving (SAE Level 2) using metrics developed based on analysis of glance behaviour.
Lessons learned to inform the conduct of trials
- There is value in trials including consideration of how trialling organisations will monitor and address human driver or operator inattention. Driver inattention (and distraction & mode confusion) during automated vehicle use poses major safety risks, and these behaviours were seen many times during our trial. This is also evident from a significant amount of research, as well as from crash investigations of injuries and fatalities that have occurred in Automated Vehicle testing and with driver assistance systems.
Lessons learned that inform future technology deployments
- Drivers quickly developed trust in (and satisfaction with) ADAS features (the trial used Tesla’s autopilot), within an hour of exposure to the system. If trust is too high it could lead to driver complacency. This is particularly problematic in the earlier stages of Level 2 ADAS deployment, where automation technology is still improving and the driver is required to be able to take over control of the vehicle at all times.
- On the test track, measures of glance behaviour (i.e., the direction that drivers looked in, moment-to-moment) revealed drivers spent a greater proportion of time looking ‘off-road’ when using the ADAS feature compared to when driving manually. When visually engaged in secondary tasks drivers took significantly longer to successfully take back control of the vehicle.
- On public roads, drivers maintained a primary focus of attention to the forward roadway both with and without the ADAS feature and tested cognitive task (potential distractions). While performing the cognitive task, their scanning behaviour became more concentrated toward the road centre in a manner consistent with ‘attentional tunneling’.
- The CANdrive research program provided evidence regarding the value that driver monitoring technology offers as a safety feature for semi-automated vehicles.
Links to project lessons learned and / or benefit assessment reports
ATRF paper: The CANdrive Program: Supporting the safer introduction of Level 2 vehicle automation
Links to any other supporting project reports
Blog article: CAN Drive – Phase 1 on Sutton Road Driver Training Centre
Last updated: May 2021.