hidden visitor counter
A-Z Index  |  Sitemap  |  Search  |  Contact
prevent.jpg
eSafety  EU Commission PReVENT Subprojects:
Print edition Print edition

Survey of filtering techniques for vehicle tracking by radar equipped platforms


Aris Polychronopoulos, Nikolaos Floudas, Angelos Amditis

ICCS, Greece

8th International Conference on Information Fusion, Philadelphia, USA
25 - 29 July 2005

 

Abstract
In modern automotive safety applications the use of radar technology seems to be a promising technique. Vehicles equipped with on board radar sensors aim at detecting moving or stationary objects in the sensor’s field of view and identifying critical collision situations. Thus, tracking is a crucial topic for efficiency improvement in such systems. The presence of non linearities in measurement space models is quite common in these architectures, i.e. the existence of radial velocity measurements in absence of lateral velocity ones generates a non-linear measurement model when Cartesian coordinate based state vector is used. In this case, Extended Kalman filter techniques are proven not to reach the desired performance, while instability is very common in practice. IMM filters could be a solution to the problem, yet not the optimal as the same models have to be used. In this paper, methods to solve these non-linearities and consequently instability effects in modeling and filtering are investigated; starting from pseudo-measurements techniques, and also including the use of the arising particle filter approaches. Particle filters have the significant property of exploiting a variety of important information available, such as constraints in target vehicles’ velocity and road geometry, inserting this into the probability distributions. At the same time the delay introduced in a system by particle filters’ calculations is a serious disadvantage comparing them with other filtering techniques. The algorithms discussed in the paper are tested by means of a simulated driving scenario in a forward looking radar architecture. Comparisons of the mean RMS estimated errors for position and velocity are presented, taking also time delay under consideration when checking the overall performance.

 

 

For more information, please contact Angelos Amditis for LATERAL SAFE or Aris Polychronopoulos for ProFusion 2


To the topTo the top

HORIZONTAL ACTIVITIES
RESPONSE 3
MAPS&ADAS
ProFusion1
ProFusion2
INSAFES
PReVAL
IP Exhibition
SAFE SPEED AND SAFE FOLLOWING
SASPENCE
WILLWARN
LATERAL SUPPORT
SAFELANE
LATERAL SAFE
INTERSECTION SAFETY
INTERSAFE
VULNERABLE ROAD USERS & COLLISION MITIGATION
APALACI
COMPOSE
UseRCams
© 2004. All rights reserved . Disclaimer This website is managed by ERTICO