Thuy-Yung Tran

Analysis of classification algorithms applied to road surface recognition

The development of remote surface recognition systems is an important step in ensuring road safety. This paper examines the performance of surface classification algorithms, used for the analysis of backscattered microwave and ultrasonic signals. The novelty of our research is the joint use of data obtained from sonar and multifrequency polarimetric radar. The results demonstrate the feasibility of reliable surface classification using the proposed methodology.

Automotive system for remote surface classification

In this paper we shall discuss a novel approach to road surface recognition, based on the analysis of backscattered microwave and ultrasonic signals. The novelty of our method is sonar and polarimetric radar data fusion, extraction of features for separate swathes of illuminated surface (segmentation), and using of multi-stage artificial neural network for surface classification. The developed system consists of 24 GHz radar and 40 kHz ultrasonic sensor. The features are extracted from backscattered signals and then the procedures of principal component analysis and supervised classification are applied to feature data. The special attention is paid to multi-stage artificial neural network which allows an overall increase in classification accuracy. The proposed technique was tested for recognition of a large number of real surfaces in different weather conditions with the average accuracy of correct classification of 95%. The obtained results thereby demonstrate that the use of proposed system architecture and statistical methods allow for reliable discrimination of various road surfaces in real conditions.

Low-THz imaging radar for outdoor applications

In this paper low-THz FMCW radar is introduced for automotive applications. Ability to produce a short-range high resolution image of the road ahead of the vehicle under any weather or terrain conditions is a considerable advancement for the automotive use. The 150 GHz FMCW radar requirements and specifications are discussed in this paper. The imaging performance of the 150 GHz radar in a real road scenario is demonstrated; the experimental results of the 150 GHz FMCW and a 30 GHz stepped frequency radar are presented and compared to each other.

Automotive surface identification system based on modular neural network architecture

The development of automotive remote surface identification system is an important step in ensuring road safety. In this paper we shall discuss a novel approach which addresses the road surface classification process. This method is based on polarimetric radar and sonar data fusion and surface identification using artificial neural network. A modular artificial neural network, which is considered in the paper, allows an overall increase in classification accuracy in the presence of a large number of surface types and a large number of signal features. We shall discuss the techniques involved and present classification results that have been achieved using modular neural network.

Comparison of pedestrian reflectivities at 24 and 300 GHz

The Radar Cross Section (RCS) is used to characterize the reflectivity of a target, which is essential for the detection performances of an automotive sensor system. Results of RCS measurements of a child mannequin at 300 GHz are presented for the first time and compared with measurements at the current automotive frequency standards (24 GHz). The RCS range and angular profile measurements of the mannequin are obtained in a controlled environment with the aid of a computer controlled turntable. The potential of low-TeraHertz (low-THz) sensors to measure additional characteristic features of pedestrians and hence enhance pedestrian recognition is also shown.

Automotive surface identification system

In this paper the practical issues of automotive surface identification system development are considering. The novelty of this work is the combining of different training algorithms, neural network structures and methods to increase the classification accuracy and avoid overfitting of real-world data. The obtained results thereby demonstrate that the use of proposed system architecture and statistical methods allow for reliable discrimination of various road surfaces in real driving condition.