Collision Avoidance in Low-Speed Maneuvering using Camera Data
Abstract: In this thesis, an automatic emergency braking is developed using camera data. The vehicle used in the development is a Volvo XC90 equipped with a prototype version of a camera-based, automated parking system. The camera detection output is a 2D version of the static environment populated with obstacles represented by convex hulls. These are used to safely avoid real-world obstacles by braking if the car is on a collision course with any obstacle. The code for analyzing the data and performing a threat assessment and decision making was written in Python. The thesis resulted in a working demo where the car successfully avoids a side collision by performing a rotation. There are some imperfections though and the function needs additional work in order to be good enough for production, particularly in sensor fusion with, e.g., ultrasonic sensors which can detect objects that the camera is struggling to detect.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)