Collision detection

The collision detection application develops the required services for warning a driver about general objects and Vulnerable Road Users (VRUs) that may cross the driving path. Exploiting the MASA infrastructure, city cameras and sensors, and data collected from a fleet of connected vehicles, the information is elaborated in real-time to detect any hazardous situations that may endanger the safety of the drivers and VRUs.

As a first step, the CLASS architecture collects and aggregates data from the MASA infrastructure and connected vehicles. By performing object detection and tracking at the edge-side, valuable information such as the type (e.g., car, bicycle, pedestrian, etc.), position, speed, etc., of the detected objects is extracted and stored in the DKB.

CLASS vehicle detecting obstacles in real time

By applying advanced data analytics on the collected information, the future trajectory of the most relevant objects (e.g., moving vehicles and VRUs) is calculated, enabling the detection of potential collisions. An alert is, then, generated and forwarded to the involved connected vehicles, increasing the driver’s awareness and giving them a 2-second margin to respond to the critical situation.

The implementation of such a distributed awareness protocol in the monitored area enables the communication of potential risks to the drivers, improving driving safety, especially in case of blind spots that may occur in intersections and parking lots.