Dynamic Obstacle Detection and Avoidance for Autonomous Driving Vehicles

Work Description:

Back to CURNIS Giovanni page
Back to USAD, Urban Shuttles Autonomously Driven

Currently, we have a Pedestrian Detection module (stage Capra Francesco), which detects dynamic obstacles from the images that come from a stereoscopic rig mounted on the USAD vehicle (pedestrian detection) and we have a costmap2d plugin (stage RADICE Carlo) that interfaces with the local planer and it's able to represent static and dynamic obstacle on a 2d map (costmap2d).

The goal is to establish the communication between the module and the plugin, so when the Pedestrian Detection module recognizes an obstacle, like a pedestrian or a car, it sends a notifications to the plugin. The plugin is the also able to make a prediction of the obstacle motion (timed trajectory); this prediction allows the car to change its route to avoid the obstacle, by means of the planner intervention.