5G-Connected Remote-Controlled Semi-Autonomous Car Trial

Abstract

This paper presents a few observations from our 5G connected remote-controlled car with semi-autonomous features. The paper focuses on presenting the framework and typical data flow used to control the car's features through a video-assisted remote driving system. The remote unit (Simulator) was connected to the 5G network to communicate with the car through a 5G radio Base-station. A 5G UE radio was fit into the car and used as a modem to interface with the car's control system. In this trial, the focus was to control the accelerator, steering, and braking mechanisms remotely. The remote unit had a human interface through a gaming console consisting of steering, braking, and accelerator pedals. These elements complete integrations were based on the Robot Operating System (ROS) to deliver commands within than 10ms latency for proper operations. The car had a programmable digital interface using CAN bus. The remote unit displayed video streams from the car's front and back view to assist the driving. The paper details the messaging architecture and measurements from the trial and compares the results with a WiFi-based modem. These results are initial measurements, and more optimization could be possible.