Development of an Autonomous Reverse Engineering Capability for Controller Area Network Messages to Support Autonomous Control Retrofits

TL;DR

This research develops a method to reverse engineer vehicle CAN messages related to accelerator and brake controls without prior vehicle knowledge, aiding the development of aftermarket autonomous vehicle retrofit kits.

Contribution

A novel algorithmic approach that identifies CAN channels for vehicle controls using IMU and CAN data, applicable across different vehicle manufacturers and standards.

Findings

Successfully identified CAN channels in six vehicles from three manufacturers.

Demonstrated potential for real-time application in aftermarket autonomous vehicle kits.

System remains effective despite changes in CAN standards.

Abstract

As the autonomous vehicle industry continues to grow, various companies are exploring the use of aftermarket kits to retrofit existing vehicles with semi-autonomous capabilities. However, differences in implementation of the controller area network (CAN) used by each vehicle manufacturer poses a significant challenge to achieving large-scale implementation of retrofits. To address this challenge, this research proposes a method for reverse engineering the CAN channels associated with a vehicle's accelerator and brake pedals, without any prior knowledge of the vehicle. By simultaneously recording inertial measurement unit (IMU) and CAN data during vehicle operation, the proposed algorithms can identify the CAN channels that correspond to each control. During testing of six vehicles from three manufacturers, the proposed method was shown to successfully identify the CAN channels for the accelerator pedal and brake pedal for each vehicle tested. These promising results demonstrate the potential for using this approach for developing aftermarket autonomous vehicle kits - potentially with additional research to facilitate real-time use. Notably, the proposed system has the potential to maintain its effectiveness despite changes in vehicle CAN standards, and it could potentially be adapted to function with any vehicle communications medium.