Gaussian mixture approach to decision making for automotive collision warning systems

Seul Ki Han, Won Sang Ra, Ick Ho Whang, Jin Bae Park

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

This paper proposes a practical probabilistic approach to collision decision making which is necessary for advanced automotive collision warning system (CWS) using FMCW radar. Most decision making algorithms assess the probable collisions based on the predicted collision position which is usually expressed as a nonlinear function of threat vehicle’s position and velocity provided by FMCW radar. Since the predicted collision position has highly nonlinear statistics in general, it is one of main obstacles to improving the reliability of the collision probability computation and to developing real-time decision making algorithms. This motivates us to devise a Gaussian mixture method for collision probability calculation with the help of linear recursive time-to-collision (TTC) estimation. The suggested TTC estimator provides an accurate TTC estimate with small estimation error variance hence it enables us to approximate the probability density function of the predicted collision position as the weighted sum of just a few Gaussian distributions. Therefore, our approach could drastically reduce the inherent nonlinearity of collision decision making problem and computational complexity in collision probability calculation. Through the simulations for the typical engagement scenarios between the host and threat vehicles, the performance and effectiveness of the proposed algorithm is compared to those of the existing ones which require heavy computational burden.

Original languageEnglish
Pages (from-to)1182-1192
Number of pages11
JournalInternational Journal of Control, Automation and Systems
Volume13
Issue number5
DOIs
Publication statusPublished - 2015 Oct 29

Fingerprint

Alarm systems
Decision making
Radar
Gaussian distribution
Error analysis
Probability density function
Computational complexity
Statistics

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Computer Science Applications

Cite this

@article{52db0093ac21459796b423df27458509,
title = "Gaussian mixture approach to decision making for automotive collision warning systems",
abstract = "This paper proposes a practical probabilistic approach to collision decision making which is necessary for advanced automotive collision warning system (CWS) using FMCW radar. Most decision making algorithms assess the probable collisions based on the predicted collision position which is usually expressed as a nonlinear function of threat vehicle’s position and velocity provided by FMCW radar. Since the predicted collision position has highly nonlinear statistics in general, it is one of main obstacles to improving the reliability of the collision probability computation and to developing real-time decision making algorithms. This motivates us to devise a Gaussian mixture method for collision probability calculation with the help of linear recursive time-to-collision (TTC) estimation. The suggested TTC estimator provides an accurate TTC estimate with small estimation error variance hence it enables us to approximate the probability density function of the predicted collision position as the weighted sum of just a few Gaussian distributions. Therefore, our approach could drastically reduce the inherent nonlinearity of collision decision making problem and computational complexity in collision probability calculation. Through the simulations for the typical engagement scenarios between the host and threat vehicles, the performance and effectiveness of the proposed algorithm is compared to those of the existing ones which require heavy computational burden.",
author = "Han, {Seul Ki} and Ra, {Won Sang} and Whang, {Ick Ho} and Park, {Jin Bae}",
year = "2015",
month = "10",
day = "29",
doi = "10.1007/s12555-014-0238-6",
language = "English",
volume = "13",
pages = "1182--1192",
journal = "International Journal of Control, Automation and Systems",
issn = "1598-6446",
publisher = "Institute of Control, Robotics and Systems",
number = "5",

}

Gaussian mixture approach to decision making for automotive collision warning systems. / Han, Seul Ki; Ra, Won Sang; Whang, Ick Ho; Park, Jin Bae.

In: International Journal of Control, Automation and Systems, Vol. 13, No. 5, 29.10.2015, p. 1182-1192.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Gaussian mixture approach to decision making for automotive collision warning systems

AU - Han, Seul Ki

AU - Ra, Won Sang

AU - Whang, Ick Ho

AU - Park, Jin Bae

PY - 2015/10/29

Y1 - 2015/10/29

N2 - This paper proposes a practical probabilistic approach to collision decision making which is necessary for advanced automotive collision warning system (CWS) using FMCW radar. Most decision making algorithms assess the probable collisions based on the predicted collision position which is usually expressed as a nonlinear function of threat vehicle’s position and velocity provided by FMCW radar. Since the predicted collision position has highly nonlinear statistics in general, it is one of main obstacles to improving the reliability of the collision probability computation and to developing real-time decision making algorithms. This motivates us to devise a Gaussian mixture method for collision probability calculation with the help of linear recursive time-to-collision (TTC) estimation. The suggested TTC estimator provides an accurate TTC estimate with small estimation error variance hence it enables us to approximate the probability density function of the predicted collision position as the weighted sum of just a few Gaussian distributions. Therefore, our approach could drastically reduce the inherent nonlinearity of collision decision making problem and computational complexity in collision probability calculation. Through the simulations for the typical engagement scenarios between the host and threat vehicles, the performance and effectiveness of the proposed algorithm is compared to those of the existing ones which require heavy computational burden.

AB - This paper proposes a practical probabilistic approach to collision decision making which is necessary for advanced automotive collision warning system (CWS) using FMCW radar. Most decision making algorithms assess the probable collisions based on the predicted collision position which is usually expressed as a nonlinear function of threat vehicle’s position and velocity provided by FMCW radar. Since the predicted collision position has highly nonlinear statistics in general, it is one of main obstacles to improving the reliability of the collision probability computation and to developing real-time decision making algorithms. This motivates us to devise a Gaussian mixture method for collision probability calculation with the help of linear recursive time-to-collision (TTC) estimation. The suggested TTC estimator provides an accurate TTC estimate with small estimation error variance hence it enables us to approximate the probability density function of the predicted collision position as the weighted sum of just a few Gaussian distributions. Therefore, our approach could drastically reduce the inherent nonlinearity of collision decision making problem and computational complexity in collision probability calculation. Through the simulations for the typical engagement scenarios between the host and threat vehicles, the performance and effectiveness of the proposed algorithm is compared to those of the existing ones which require heavy computational burden.

UR - http://www.scopus.com/inward/record.url?scp=84942456252&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84942456252&partnerID=8YFLogxK

U2 - 10.1007/s12555-014-0238-6

DO - 10.1007/s12555-014-0238-6

M3 - Article

AN - SCOPUS:84942456252

VL - 13

SP - 1182

EP - 1192

JO - International Journal of Control, Automation and Systems

JF - International Journal of Control, Automation and Systems

SN - 1598-6446

IS - 5

ER -