LaonRoad Pedestrian Collision Avoidance System provide real-time pedestrian and driver collision alarms by EDGE-based AI video detection.
It receives traffic signal information from the CVIB (Connected Vehicles I/F Board) or option board of the on-site traffic signal controller directly, and it can provide pedestrian collision alarms directly to connected vehicles and autonomous vehicles through the roadside base station (RSE).
LaonRoad C-ITS Pedestrian Collision Avoidance System provides a stable system based on experiences in the Pangyo Autonomous Driving Center pilot project and the Gwangju C-ITS demonstration project.
EDGE-based AI Video Detection * The Pedestrian Collision Avoidance System of C-ITS must propagate the collision situation between vehicles and pedestrians very quickly, but in the case of a center-based system, it’s difficult to guarantee real-time performance due to delay when transmit the video. * LaonRoad LPGR-100C is the first camera-integrated EDGE device developed in Korea, which guarantees real-time performance by transmitting the information to the RSU or vehicle immediately with deep learning calculation (10 times per second). |
|
Vehicle and pedestrian detection using deep learning * Pedestrians and cars are detected using a lightweight deep learning network optimized for pedestrian collision avoidance and EDGE devices. * Events are extracted with considering moving direction of vehicle and pedestrian. * Events are extracted in connecting with traffic signal information. - Normal pedestrians are displayed in green, jaywalkers are displayed in red - Detects the vehicles that violate the stop line |
Gwangju City C-ITS Pedestrian Collision Avoidance System Deep Learning Analysis Demo |
V2X-based pedestrian/vehicle collision avoidance warning service * Provide a safe walking system based on C-ITS by transmitting pedestrian collision avoidance alarm information which extracted from AI video detector to connected vehicle (CV) and autonomous vehicle (AV) with V2X communication through RSE (roadside base station). |
|