International Symposium on Antennas and Propagation


Session Number:3D07



Collision-Avoidance Algorithm with High Precision on Location, Velocity and Acceleration

Po-Jen Tu,  Jean-Fu Kiang,  Chia-Cheng Ho,  


Publication Date:2008/10/27

Online ISSN:2188-5079


PDF download (242.9KB)

Adaptive cruise control (ACC), parking aid, collision avoidance, and pre-crash detection require reliable calculation pending on the accurate estimation of target location, velocity and acceleration. Several methods have been proposed to trilaterate a specific target based on detecting these kinetic parameters from multiple signals. In [1], Klotz presents a method to trilaterate a target using multiple FMCW echoes in a least-square sense. Given the relative range measured at the sensor, a one-stage linear Kalman filter is used to estimate the relative range and relative velocity [2]. However, the acceleration can not be estimated accurately when the target is making a turn in a short period. Furthermore, a two-stage linear Kalman filter has been built upon the one-stage linear Kalman filter, with a new bias vector to estimate the relative acceleration [3]. Nevertheless, the convergence time will be too long if the target is making a turn in a short period. Hence, the extended Kalman filter is used to estimate all the kinetic parameters of the target using only one sensor [4]. However, the error in some parameters can be very large when the target is making a turn. In this work, we propose a minimum-delay Kalman filter to reckon the kinetic paramaters by taking the measured FMCW (frequency-modulated continuous wave) echoes. A second Kalman filter is also used after trilateration to significantly reduce the convergence time for estimating the acceleration. Demonstration is given with a vehicle making a turn across the adjacent lane in front of the host vehicle. The trajectory of moving target with respect to the host can be calculated to predict the time for collision.