Overall goal: Design and develop a perception-based self-driving system for urban scenarios, performing a demonstration at the UAB.
Achieving this overall goal, involves reaching intermediate goals:
- Design and develop navigation software based on the use of vision, radar, GPS/IMU, vehicle state, and maps.
- Design and develop object detection software specialized in dynamic traffic participants (non-cooperative vehicles, pedestrians, and bicyclists), with the capability of anticipating their intentions (for a quick reaction in dangerous situations). This software will rely on vision, radar, vehicle state, and ultrasound data.
- Design and develop software to detect unknown obstacles lying on the road (hazards).
- Design and develop software for detecting road marks (lane markings, traffic signs, zebra crossings, and semaphores), in order to respect traffic rules, using vision.
- Design and collect representative data sets for benchmarking the self-driving capabilities.
- Design and develop a simulation platform that can playback the benchmarking data and evaluate the vehicle’s behavior off-line.
- Design and execute self-driving demonstrations in controlled urban areas (in the UAB campus of Bellaterra).
- Incorporating cooperative information into the self-driving procedure by collaborating with other groups.