Researchers have developed technology that allows autonomous vehicles to track pedestrians and cyclists hidden from the line of sight by obstacles like buildings
Spotted: Researchers at the University of Sydney’s Australian Centre for Field Robotics have developed technology that can allow autonomous vehicles to ‘see’ pedestrians and cyclists who are located in blind spots or obscured by fast-moving vehicles. The research was delivered in collaboration with connected vehicle company Cohda Wireless and the iMOVE Cooperative Research Centre.
The technology uses roadside information-sharing units, equipped with additional sensors such as cameras and lidar, that allow autonomous vehicles to share what they ‘see’ with each other using vehicle-to-vehicle communication. This system allows the vehicles to access a number of different viewpoints, significantly increasing their range of perception and allowing the connected vehicles to see things they wouldn’t normally.
Tests of the system demonstrated that vehicles were able to predict potential threats to safety, such as a pedestrian rushing towards a crossing area. The vehicles could then take pre-emptive action, for example braking and stopping before the pedestrian reached the crossing area. The technology, called cooperative or collective perception (CP), is being commercialised by Cohda, and the engineers developing the technology hope that it could eventually be used to benefit all vehicles, even those not connected to the system.
Professor Eduardo Nebot, from the Australian Centre for Field Robotics, described the system as a, ‘game changer for both human-operated and autonomous vehicles’. He added that, “The connected vehicle was able to track a pedestrian visually obstructed by a building with CP information. This was achieved seconds before its local perception sensors or the driver could possibly see the same pedestrian around the corner, providing extra time for the driver or the navigation stack to react to this safety hazard”.
Autonomous vehicle technology is moving ahead by leaps and bounds, but it still suffers from some significant limitations – including a limited ability to predict what people or animals might do. The work by the Australian team may go some way to overcoming this, along with recent innovations such as remote-piloted ride sharing and in-car radar warning systems.
Written By: Lisa Magloff