New technology developed by Australian researchers could help autonomous vehicles avoid pedestrians and cyclists with no line of sight, according to researchers of University researchers.
The team of the Australian Center for Field Robotics at the University of Sydney reported that they were able to dramatically improve detection capability by linking vehicles to networks of road sensors and to each other, resulting in what they called ‘cooperative or collective perception (CP) “. The “intelligent road units” (IRSU) are equipped with equipment such as lidar sensors or cameras, which can then share the information with passing vehicles, which in turn share it with others connected to the same network. The researchers compared the overall effect to an x-ray, allowing cars to be aware of objects falling out of line of sight (such as a pedestrian behind a building, or a cyclist hidden by another vehicle).
While engineering difficulties have been the main obstacle to the development of self-driving cars, to resolve Safety concerns would also be required for them to receive regulatory approval or reassure a public it remains anxious on handing the flywheel back to machines. Some earlier studies on the detection of hidden dangers relied on technologies such as laser sensors and x-rays, or in the case of a MIT Report 2019, real-time lighting and shadow analysis that can detect the approach of a pedestrian or a vehicle. The approach detailed in Australian research provides more data supporting an alternative approach, combining communication between cars with smart roads to offer each vehicle several points of view.
According to the team report, vehicle-to-X communication (V2X) of environmental awareness data can “be a game-changer for human-controlled vehicles and autonomous vehicles”, not only because the increasing use and standardization of these systems link equipped vehicles, but allow them to share warnings from other objects on the road that are not. They also discussed how such a collective perception system could be designed to allow connected vehicles to account for uncertainties, such as sensors that are not 100% accurate or subject to noise and environmental interference, or how accurately a car follows its own position on the road. The report also details other innovations, such as methods of distinguishing and tracking specific pedestrians.
Professor Eduardo Nebot of the Australian Field Robotics Center said in a hurry Release that researchers believe the technology could “dramatically improve the efficiency and safety of road transport.”
Researchers conducted several tests to demonstrate the potential of such a system to detect road hazards and prevent accidents. In a test involving a human-driven car in an urban environment in Chippendale, Sydney, the use of an IRSU allowed the vehicle to anticipate traffic activity “well beyond the range of its on-board perception sensors” as well as “see” a “visually obstructed pedestrian behind a building” a few seconds before he would have done so otherwise. Another test in a lab environment used an IRSU to provide a car with simulated data about a jogger heading towards an intersection, where the car successfully braked and gave way “based on the expected future condition of the vehicle. pedestrian ”before he even enters the road. Other tests involving several vehicles linked via an IRSU, as well as those carried out in the open source CARLA simulator, had equally satisfactory results and allowed vehicles to anticipate obstacles such as pedestrians and other normal cars on the road. .
“Our research has shown that a connected vehicle can ‘see’ a pedestrian on bends,” Mao Shan, the project’s lead researcher, said in the statement. “Most importantly, we demonstrate how connected autonomous vehicles can autonomously and safely interact with walking and running pedestrians, relying solely on information from the ITS bus station. “
The main obstacle for the system detailed in the report is that it would require investment in infrastructure to build roadside sensor networks that would help feed car situational awareness data. But the research team argued that it could help reduce the cost of autonomous driving and hazard detection in human-driven vehicles by sharing some of the required equipment, as well as aligning well with existing roads..
“[Collective perception] enables intelligent vehicles to break the physical and practical limits of on-board perception sensors and, in the meantime, adopt improved perception quality and robustness as well as other expected benefits of CP service and V2X communication ” , the researchers concluded in the report. “Equally important, CP can also reduce reliance on local vehicle perception information, reducing the need and cost of on-board sensing systems. “
‘In addition, it has been shown that, when properly used, perception data from other [intelligent transportation systems] can be used as another reliable source of information to add additional robustness and integrity to stand-alone operations,», They added.
According to the report, the research team plans to continue developing the system, including developing “more advanced and ready-to-deploy” platforms, creating an open standard for collective perception vehicles and testing systems. more advanced collective perception in both human-piloted and autonomous vehicles in more complex traffic environments.