Surface Recognition for Cars: A Comprehensive Approach for Neural Networks

Abstract

This paper explores the viability of neural-network-based classification of ground surface for vehicles. By classifying road surface in near real-time, improvements in vehicle performance (e.g. braking and cornering) may be possible. Classification performance for many combinations of feature encoding and neural network types are compared. The vehicle used here was an Audi “S3” with a magnetic suspension system on the Sport mode. An NI CompactRIO (or cDAQ) module was used to record from a lowing the cDAQ to communicate with the PCB 352C03 one-axis accelerometer. The accelerometer was firmly attached to the windshield of the car. This work focuses on the classification of four road surfaces (asphalt, dirt, concrete, and sand), though larger target sets were also considered. The most accurate method involved a MATLAB feature extraction package with a back-propagation neural network, yielding an overall accuracy of 97%. Lessons learned from this wide exploration of options may extend to other related classification problems.

Emergency-braking Distance Prediction using Deep Learning

“…The other challenge is the robustness on various road surfaces and conditions since most FCW and AEB algorithms are designed for a single road surface under optimal conditions. Previous studies [18][19][20][21] proposed various surface classification methods for optimizing the performance of the vehicle based on the surface. Giguere and Dudek [18] designed a simple probe to identify the surface consisting of an accelerometer at the end of a bar that was dragged behind the vehicle while traveling at low speed.…”