This study used data from 3D Inertial Measurement Unit (accelerometers/gyroscopes) mounted on the Powered Two Wheelers (PTW) to analyze and classify PTW rider’s behavior. In our work, we hypothesize that by learning riding patterns, useful information pertaining to the conditions of riding environment can be provided for riders in order to assist them on the one hand, and on the other hand, we can provide efficient tools that can be exploited in a more general aim of designing expert systems to aid the researchers in performing an in-depth study of PTW rider’s activity rider’s activity in these specific situations. Therefore, the riding pattern recognition problem is formulated as a classification problem aiming to identify the class of the riding situation namely, Straight Line (SL), Right Turn (RT), Left Turn (LT), Round About (RA) and STop (ST). A key issue in the design of such systems is that in the first stage, six machine learning techniques are used as single classifiers and in the second stage these classifiers are combined to constitute a multi-expert system. Two main strategies are applied for combining classifiers: fusion based on majority voting rule and fusion based on Bayesian formalism.