End-to-End Latency Measurement Methodology for Connected and Autonomous Vehicle Teleoperation

TL;DR

The paper tackles the challenge of measuring end-to-end teleoperation latency in connected and autonomous vehicles by jointly quantifying motion-to-motion ($M2M$), glass-to-glass ($G2G$), and end-to-end ($E2E$) latencies. It introduces a hardware-assisted framework using two GPS-synchronized Raspberry Pi 5 units, gyroscopes, a TEPT-4400 phototransistor, and an LED in the camera FOV to capture precise timing of steering motion onset, camera capture, and vehicle actuation, with clock synchronization via GPS and Chrony. The authors provide a formal latency breakdown, baseline assessments showing millisecond-scale measurement precision, and field tests over 4G and 5G NSA networks that reveal $M2M$, $G2G$, and $E2E$ medians around 300 ms, 200 ms, and 500 ms respectively, with 5G offering modest improvements. They conclude that both network performance and hardware-induced delays (camera sampling, display refresh, and actuation) shape total latency, and propose future hardware optimizations and extended synchronization to achieve finer decomposition and reduced M2M error. Overall, the framework enables accurate, hardware-grounded latency diagnostics essential for optimizing teleoperation in real-world CAV deployments.

Abstract

Connected and Autonomous Vehicles (CAVs) continue to evolve rapidly, and system latency remains one of their most critical performance parameters, particularly when vehicles are operated remotely. Existing latency-assessment methodologies focus predominantly on Glass-to-Glass (G2G) latency, defined as the delay between an event occurring in the operational environment, its capture by a camera, and its subsequent display to the remote operator. However, G2G latency accounts for only one component of the total delay experienced by the driver. The complementary component, Motion-to-Motion (M2M) latency, represents the delay between the initiation of a control input by the remote driver and the corresponding physical actuation by the vehicle. Together, M2M and G2G constitute the overall End-to-End (E2E) latency. This paper introduces a measurement framework capable of quantifying M2M, G2G, and E2E latencies using gyroscopes, a phototransistor, and two GPS-synchronized Raspberry Pi 5 units. The system employs low-pass filtering and threshold-based detection to identify steering-wheel motion on both the remote operator and vehicle sides. An interrupt is generated when the phototransistor detects the activation of an LED positioned within the camera's Field Of View (FOV). Initial measurements obtained from our teleoperated prototype vehicle over commercial 4G and 5G networks indicate an average E2E latency of approximately 500 ms (measurement precision +/- 4 ms). The M2M latency contributes up to 60% of this value.

Figures (5)

• Figure 1: Measurement framework architecture, running on two GPS-synchronized rpi 5, using two MPU-6050 for motion detection and TEPT4400 phototransistor for camera event detection.
• Figure 2: Relationships between the latency components and primary breakdown of m2m and g2g.
• Figure 3: Experimental setup for latency measurement framework field-tests.
• Figure 4: m2m, g2g, and e2e latency measurements using a 4G network compared with a 5G nsa network in a static scenario.
• Figure 5: Motion-to-Motion and Glass-to-Glass breakdown using the 5G nsa network.