Written by Kelly Daize, Strategic Market Director, Area X.O
Self-driving cars will eventually be far safer than those driven by people. That’s the promise of self-driving technology. We know we can get there, but there’s a lot of work required before that happens. The problem is large and complex, and many areas need more research to help advance the field.
To this end, Area X.O and organizations across industry, government, and academia are collaborating on a V2X project on how to improve safety for vulnerable road users at and near intersections. Part of Transport Canada’s Enhanced Road Safety Transfer Payment Program (ERSTPP), this project received funding last year to kick off its research. The nine-month research phase was recently completed; a formal report is being prepared for imminent release while the preliminary results are to be discussed December 5th at CAV Canada 2022.
Why was this project needed, what was its goal, what was done, and what was discovered?
Why did we initiate this project?
Accident-free operation is the promise for connected autonomous vehicles (CAV) and this project is looking at how V2X can assist in providing pre-emptive information to CAVs that will enable zero accidents.
Vulnerable road users (VRUs) are pedestrians (including children), cyclists, motorcyclists, and persons with disabilities or reduced mobility and orientation, including those using personal mobilized devices (such as motorized wheelchairs and scooters). These individuals have a high risk of injury or fatality in a vehicular interaction, and a significant proportion of these accidents are encountered at intersections.
For example, nearly three out of four bicycle deaths come from vehicle collisions. There is a serious need to reduce the frequency and severity of these encounters to increase the safety of everyone in our communities.
While CAVs have tremendous potential to do this, they still need data within a reasonable time frame to be successful. Providing CAVs with information on potential VRUs well ahead of when the vehicle can detect them, may be the key to and is the impetus for our project.
We also worked with Canadian companies like Smart Cone and Fortran Traffic to deploy their innovative solutions that provide notifications through apps on cell phones and tablets.
Who was involved?
Assessing the performance, safety, and potential impact of V2X solutions around highly active intersections became our mission. Area X.O wanted to use its experience and unique private and public infrastructure to help investigate the problem with other parties that are also concerned about improving VRU safety.
To this end, we formed a diverse coalition across multiple sectors that bought Area X.O, operated by Invest Ottawa together with: startups (Cheetah Networks, Fortran Traffic, Auto Guardian by SmartCone Technologies), multinationals (BlackBerry QNX, Hexagon AutonomouStuff, Siemens), government (National Research Council Canada, City of Ottawa), and international partners engaged in the Global Initiative For Certifiable AV Safety (World Economic Forum, Deepen AI and WMG, University of Warwick).
What was our goal?
Our goal was to assess and create data-driven insights about smart mobility technologies detecting vulnerable road users at intersections. Very specifically, we wanted to understand if machine vision systems could help us avoid collisions with pedestrians or cyclists.
Are we able to lengthen the time to collision enough so that it could be avoided altogether? Can the intersection detect and transmit the presence of vulnerable users to a car with sufficient time to brake? Can we improve the car’s ability to detect a VRU in poor visibility conditions, such as nighttime or bad weather?
We wanted the information we collected to be useful in establishing best practices, policies, and regulations that could enhance the safety both of VRUs and of the drivers and passengers within connected and autonomous vehicles (CAVs).
What did we do?
We executed a wide array of rigorous tests safely and securely at Area X.O’s private R&D complex by outfitting intersections with lidar, cameras, thermal and radar sensors and roadside units that analyze the data and broadcast the safety messages to the integrated on board unit inside our CAV.
We utilize our suite of motorized test mannequins to simulate high risk VRU scenarios and utilize an integrated RF spectrum analyzer to look at the timing and distance that the vehicle received the safety information. Once this was complete, we moved this testing from our controlled campus to a public intersection in the city of Ottawa.
What did we discover?
We received an answer to our primary question about whether sensors at intersections can communicate about VRUs to cars with enough warning to do something about it. Indeed, we proved that giving road-side units the ability to detect and communicate the presence of a VRU results in quicker reactions than when car sensors to do all the work. This would give the car itself (via an autonomous system) an active safety system (automatic braking) or a passive safety system (driver alerts) additional precious seconds to avoid a disaster.
Smart mobility technologies (such as machine vision solutions, V2X technologies, deep learning-based perception systems and VRU communications) can enhance the safety of intersections, roads, and vulnerable road users. Data that has been collected from the VRU test scenarios, will be shared with the World Economic Forum and WMG Safety Pool™, a repository for CAV test scenarios. International collaboration is required if we are going to make zero collisions a reality.
Further projects are extending this work to make railroad crossings safer too.
To learn what sensors worked best, how much warning time we estimate is reliable, how weather conditions can impact the results, what protocols were used, and how communication can be best authenticated – you’ll either need to read the full report or talk to me at CAV Canada in a few short weeks!