Combining a Universal OBD-II Module with Deep Learning to Develop an Eco-Driving Analysis System

Abstract

Vehicle technology development drives economic development but also causes severe mobile pollution sources. Eco-driving is an effective driving strategy for solving air pollution and achieving driving safety. The on-board diagnostics II (OBD-II) module is a common monitoring tool used to acquire sensing data from in-vehicle electronic control units. However, different vehicle models use different controller area network (CAN) standards, resulting in communication difficulties; however, relevant literature has not discussed compatibility problems. The present study researched and developed the universal OBD-II module, adopted deep learning methods to evaluate fuel consumption, and proposed an intuitive driving graphic user interface design. In addition to using the universal module to obtain data on different CAN standards, this study used deep learning methods to analyze the fuel consumption of three vehicles of different brands on various road conditions. The accuracy was over 96%, thus validating the practicability of the developed system. This system will greatly benefit future applications that employ OBD-II to collect various types of driving data from different car models. For example, it can be implemented for achieving eco-driving in bus driver training. The developed system outperforms those proposed by previous research regarding its completeness and universality.

Keywords

• eco-driving
• universal on-board diagnostics II (OBD-II)
• controller area network (CAN)
• deep learning
• fuel consumption