Automatic parking systems consist of three core technologies: 1) target position designation; 2) path planning; and 3) path tracking. Target position-designation methods can be divided into four categories: 1) user-interface based; 2) parking slot marking based; 3) free-space based; and 4) infrastructure based. Considering the fact that parking-assist systems are expected to be used mainly in urban situations, recognition of parking slot markings could be the most economical and efficient solution for target position designation. This paper proposes a semiautomatic parking slot marking-based target position-designation method. The user can initiate parking slot marking recognition by placing a finger on each side of the entrance of the target parking slot. With such a user interface, these systems can reduce the search range to so small an area that computational loads and false-recognition rates are significantly reduced. Furthermore, by identifying the junction patterns of parking slot markings around the designated point with a neural network-based classifier, the user can establish the target position with a uniform user interface. The proposed system showed a 91.10% recognition rate in 191 test cases consisting of five different types of parking slot markings.
Bibliographical noteFunding Information:
Manuscript received April 10, 2009; revised August 25, 2009 and October 1, 2009. First published October 20, 2009; current version published February 19, 2010. This work was supported by the Korea Science and Engineering Foundation under Grant R112002105070010(2009) through the Biometrics Engineering Research Center, Yonsei University. The review of this paper was coordinated by Dr. Y. Gao.
All Science Journal Classification (ASJC) codes
- Automotive Engineering
- Aerospace Engineering
- Electrical and Electronic Engineering
- Applied Mathematics