Expanding Operating Design Domain of automated vehicles

With regards to automated vehicles, governments in Australia have a regulatory role and an infrastructure manager role. They need to understand the challenges associated with unique Australian uses cases as a regulator and also have a role to play in providing information to automated vehicles in a timely manner as an infrastructure manager.

Through work package 3 of the iMOVE project 1-008, Cooperative and Highly Automated Driving Safety Study (CHAD), six use cases have been identified as unique to Australia, these are:

1. Road trains
2. Australian unique wildlife
3. Narrow regional roads with two-way traffic
4. Flooded roads
5. Intersections with staggered stop lines
6. Different types of lane markings such as wide centre lines

This project seeks to resolve three of the six use cases and share the findings.

Participants

• Department of Transport and Main Roads (Queensland)
• Queensland University of Technology
• CARRS-Q

Project background

The CHAD project investigated safety benefits and impacts of highly automated vehicles. Work Package 3 (WP3) of the CHAD pilot identified 13 priority use cases where automated vehicles (AV) may require specific development to cater for Australian unique conditions.

After identifying these use cases, international experts were consulted to evaluate their relevance, along three criteria: Australia Specific, Safety Benefit and Technical Feasibility. The ranking was used to prioritise use cases to be investigated given the budget and timing of the CHAD project. The international experts agreed on three use cases as unique to Australia and identified a further three use cases which are not unique to Australia but have higher exposure in Australia.

The three use cases identified as unique to Australia are:

• Road trains
• Australian unique wildlife
• Narrow regional roads

The three cases not unique to Australia but with higher exposure in Australia are:

• Flooded roads
• Intersections with staggered stop lines (visibility issues)
• Different types of lane markings (such as wide centre lines)

CHAD WP3’s scope was limited to resolving just one out of the above first three use cases. Due to the limit in scope the CHAD team selected the road train use case as it had sufficient data to resolve this use case. This meant the remaining five use cases and related issues were not investigated as part of CHAD.

These use cases provide safety benefits and a broader understanding of the operation of automated vehicles on Australian roads.

Research need

Governments in Australia have a regulatory role and an infrastructure manager role. They need to understand the challenges associated with these use cases as a regulator and also have a role to play in providing information to automated vehicles in a timely manner as an infrastructure manager. This could be through clear road signs/markings, HD maps or Cooperative Intelligent Transport Systems (C-ITS) functions.

To accelerate AV adoption in Australia, this project seeks to resolve three of the remaining five use cases and share the learnings primarily for the government as regulator and infrastructure manager. Three use cases will be selected based upon availability of adequate data.

The challenges associated with each of the five use cases are detailed below:

1. Australian unique wildlife: Initial data recording for this use case during the CHAD 1-008 project was unsuccessful mainly due to ZOE2 research vehicle insurer requirements. There are perceived challenges in detecting, classifying, and predicting future trajectory of Australian unique wildlife.
2. Narrow regional roads: Regional roads with about 4 metre sealed surface with bi-directional traffic is considered to be difficult for AVs to deal with. This use case further becomes challenging as passing manoeuvre creates a dusty environment.
3. Flooded roads: no research has been conducted on how a vehicle’s sensors can detect and understand flooded roads and associated intensity and react accordingly.
4. Intersections with staggered stop lines: Intersections with visibility issues when an AV stops at the intersection with staggered stop lines are a particular challenge. Situations such as this become more challenging when maps are not accurate. Furthermore, path planning on an intersection under differing environmental conditions can create further complexity.
5. Different types of lane markings: Unique lane markings such as wide centre lines on Queensland’s highways may create challenges for algorithms which are developed outside Australia.

In summary, very little is known about how in-vehicle sensors perceive and understand such events, how the automated vehicle should react, re-plan and what the optimal way to transmit the missing information (if any) to the vehicle.

It is critical to understand the safety risk associated with these situations and inform regulator and infrastructure providers to facilitate early adoption of AVs.

Project intent

This project seeks to collect data on the public roads and for some use cases establish capabilities at the RACQ Mobility Centre of Excellence such as road flooding infrastructure, installing associated C-ITS stations and reconfiguring a C-ITS equipped signalised intersection.

Following the development of these facilities, or public road data collection, the project will select three of the above mentioned five scenarios and provide recommendations on how AVs can detect, understand, and negotiate them safely.

It is critical to ensure the test track environment matches the real road environment. It is expected that in some of these use cases, the infrastructure manager may need to provide additional information to AVs either via C-ITS or HD maps.

QUT’s ZOE2 vehicle will be extensively used to test these scenarios and may be extended with other detection devices to improve the understanding of the sensor’s limitations.

Project objectives

The objectives of this project are to:

1. Create a knowledge base for the regulatory authorities.
2. Understand what additional support an infrastructure manager may need to provide for expanding the Operating Design Domain (ODD) of L3/L4 automated vehicles.

Please note …

This page will be a living record of this project. As it matures, hits milestones, etc., we’ll continue to add information, links, images, interviews and more. Watch this space!