Hough Transform for Parking Space Estimation Using Long Range Ultrasonic Sensors 2006-01-0810
This paper describes a system for supporting the driver of a passenger car in different parking situations. Todays cars are getting larger in size and the drivers view in both forward and rearward direction is becoming more limited. This fact calls for a system of sensors and algorithms capable of supporting the driver through the parking maneuvre in a safe and smooth way.
The paper presents the development of some of the subsystems in a fully automatic parallel parking system, utilizing ultrasonic ranging sensors for environment mapping. In contrast to existing passive parking aid systems, the ultrasonic range sensors need to have a narrower aperture to be able to map the surroundings properly. This can be accomplished by either increased sensor size or by a higher number of sensors.
The emphasis of the paper is the signal conditioning in the parking system. The Hough-transform along with a statistical CUSUM test are used to find the properties of the target parking space.
The system makes use of hardware components already available in modern cars and a small number of added components. The resulting system is automated, from finding a suitable parking space to maneuvering the car into the parking space, while keeping the driver in authority since the longitudinal control, i.e. throttle and brakes, are still the drivers responsibilities.
The prototype system is implemented in a Volvo S60 which has been modified with an electric power steering unit and an ultrasonic sensor array consisting of a total of six sensors spread out around the vehicle. The electric power steering unit is used for steering wheel angle control when the system is active by adding an external torque to the assist torque normally applied.