This paper addresses the detection of wheel faults in autonomous vehicles. Instead of the typically broad range of sensors involved, localization data is used to detect and classify three major faults in torque-controlled DC motors. A four wheeled vehicle is implemented in simulation with independent steering and in-hub motors to generate localization data. The vehicle model is based on extensive vehicle dynamics modeling to accurately predict a small passenger vehicle. These three faults are induced on the vehicle to determine the effectiveness of the localization method and test its ability to detect the faults and delineate between the different fault types. Lastly, an extension is outlined for detection and classification for broader error types beyond those represented by the three errors examined.