Context-sensitive bicycle and pedestrian detection and tracking

The detection and tracking of bicycles and pedestrians is an important technology to pursue for the sake of automotive safety for both manual
and autonomous cars. We are developing algorithms which fuse cameras, lidar, and radar to detect pedestrians on foot as well as on bicycles on all sides of a vehicle. In particular, we are focusing on detecting child-sized pedestrians and people that are not necessarily standing but are in other body poses that are not addressed by currently existing camera systems such as people on crutches, using walkers, or in wheelchairs. Different sensor technologies and types are being evaluated to determine their strengths and weaknesses
(e.g. performance vs. cost) for the different domains. Wherever possible, contexts of the surrounding world model will be used to improve detection and tracking.