Localization for Automatic Parking System using Interacting Multiple Model Kalman Filter with Rear-View Camera

摘要

This paper presents a new vehicle localization method that improves rear-view camera usage during automatic parking. Measuring the position of a parking stall with cameras is significantly affected by vehicle's pitch motion during parking. The pitch motion is influenced by the number of passengers and its jerk while moving backward, and leads to inaccurate distance measurement during parking. To cope with the problem, a new vehicle localization using an interacting multiple model (IMM) Kalman filter is proposed. The IMM filter is to improve the accuracy of the camera measurement influenced by pitch changes during backward motion. In this paper, we used two models based on the car's pitch motion. We obtained improved vehicle localization performance using the IMM Kalman filter and validated its performance experimentally via a real-vehicle experiment with a various number of passengers.