Low Cost, Open-Source Testbed to Enable Full-Sized Automated Vehicle Research

TL;DR

This paper introduces a low-cost, open-source vehicle testbed based on ROS for research in vehicle automation, demonstrating remote control, system identification, and path following with experimental validation.

Contribution

It presents a novel, accessible testbed platform for vehicle automation research, including hardware/software design, control methods, and validation techniques.

Findings

Successfully controlled a full-sized vehicle remotely

Developed a differentially flat system for vehicle control

Validated a path following algorithm experimentally

Abstract

An open-source vehicle testbed to enable the exploration of automation technologies for road vehicles is presented. The platform hardware and software, based on the Robot Operating System (ROS), are detailed. Two methods are discussed for enabling the remote control of a vehicle (in this case, an electric 2013 Ford Focus). The first approach used digital filtering of Controller Area Network (CAN) messages. In the case of the test vehicle, this approach allowed for the control of acceleration from a tap-point on the CAN bus and the OBD-II port. The second approach, based on the emulation of the analog output(s) of a vehicle's accelerator pedal, brake pedal, and steering torque sensors, is more generally applicable and, in the test vehicle, allowed for the full control vehicle acceleration, braking, and steering. To demonstrate the utility of the testbed for vehicle automation research, system identification was performed on the test vehicle and speed and steering controllers were designed to allow the vehicle to follow a predetermined path. The resulting system was shown to be differentially flat, and a high level path following algorithm was developed using the differentially flat properties and state feedback. The path following algorithm is experimentally validated on the automation testbed developed in the paper.