Evaluation of estimation approaches on the quality and robustness of collision warning system

TL;DR

This paper evaluates the effectiveness of three estimation algorithms—constant velocity, constant acceleration, and Kalman filter—in improving the robustness and accuracy of collision warning systems amid data loss in vehicle communication networks.

Contribution

It compares the performance of these algorithms in estimating lost data to enhance collision warning system reliability under various data loss conditions.

Findings

Kalman estimator shows superior accuracy and robustness.

Constant velocity and acceleration methods are less effective in high data loss scenarios.

Improved data estimation enhances collision warning system performance.

Abstract

Vehicle safety is one of the most challenging aspect of future-generation autonomous and semi-autonomous vehicles. Collision warning systems (CCWs), as a proposed solution framework, can be relied as the main structure to address the issues in this area. In this framework, information plays a very important role. Each vehicle has access to its own information immediately. However, another vehicle information is available through a wireless communication. Data loss is very common issue for such communication approach. As a consequence, CCW would suffer from providing late or false detection awareness. Robust estimation of lost data is of this paper interest which its goal is to reconstruct or estimate lost network data from previous available or estimated data as close to actual values as possible under different rate of lost. In this paper, we will investigate and evaluate three different algorithms including constant velocity, constant acceleration and Kalman estimator for this purpose. We make a comparison between their performance which reveals the ability of them in term of accuracy and robustness for estimation and prediction based on previous samples which at the end affects the quality of CCW in awareness generation.