SLIP PREDICTION USING SVM

Abstract

Understanding and calculating the factors that affect the vehicle-terrain interaction is crucial inithe domains ofigroundivehiclesiandigroundimobile robotsisinceitheisuccess or failure of an entire operation depends on them. This study uses a machine-learning algorithm to solve a key problem: predicting tire slip. The performance of traditional machine learning method (i.e., Support Vector Machine (SVM)) against Deep Neural Networks (DNNs) and Convolutional Neural Networks are compared using data sets collected by ground robotic platforms (MSL Curiosity rover). This research also demonstrates how network design (DNN and CNN) and tuning parameters affect theiperformance ofithoseimethods. Thisipaperialsoiprovides a detailed description of theilessons learnediinitheiimplementationiofiDNNsiandiCNNsiandihowitheseimethods might be applied to different issues.