Delay Compensation for Remote Driven Vehicles: An SRCKF-Based Predictor

Zhao, Lin; Nybacka, Mikael; Rothhämel, Malte; Mårtensson, Jonas · 2025 · OpenAlex-citations

DOI: 10.1109/tie.2025.3613626

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of communication delay in remote driving systems, which serve as backup mechanisms for automated vehicles. Network latency, encoding delays, and actuator response times degrade control stability and desynchronize motion cueing feedback, leading to a poor remote driving experience. Existing delay compensation methods are limited: model-based predictors (MBPs) require high-fidelity vehicle models and are sensitive to parameter uncertainties, while model-free predictors (MFPs) are adaptable but prone to signal overshoot and lack robustness during rapid transitions. The authors propose a Square Root Cubature Kalman Filter-based Predictor (SRCKP) to fuse MBP and MFP outputs, leveraging the structural knowledge of models and the adaptability of data-driven prediction. Additionally, the study introduces an Overshoot Compensator (OSC) to mitigate MFP distortion and a Packet Loss Predictor (PLP) to handle missing data. The proposed system architecture compensates for specific delays in the transmission loop between the Remote Control Tower (RCT) and the Remote Driven Vehicle (RDV). The SRCKP fuses state estimates from a high-fidelity vehicle dynamic model (running in the RCT using undelayed control inputs) with predictions from an enhanced 2DoF-MFP (extrapolating delayed sensor data). The vehicle model includes longitudinal, lateral, yaw, roll, and pitch dynamics, utilizing a Pacejka Magic Formula for tire forces. The OSC applies second-derivative scaling to correct overshoot in the MFP output, while the PLP uses a one-step derivative method to estimate missing packets. The system specifically targets the compensation of vehicle state transmission delays (roll, pitch, and yaw rates) to improve motion feedback fidelity. Validation was conducted through simulations and hardware-in-the-loop (HIL) experiments across two comprehensive driving scenarios. Scenario 1 involved dynamic maneuvers such as double lane changes, slalom, and large turns to test roll and yaw dynamics. Scenario 2 evaluated robustness under road disturbances, including low-friction surfaces, slopes, and banking. The results demonstrate that the SRCKP significantly outperforms traditional MFPs. In simulation, the SRCKP reduced the L2-norm error by up to 81.2% under best-case conditions. In HIL experiments, the error reduction reached up to 54.0%. The method effectively mitigated overshoot issues associated with MFPs and maintained stability despite packet loss and varying road conditions. The significance of this work lies in providing a unified, real-time delay compensation framework tailored for remote driving feedback. By fusing model-based and model-free approaches within a probabilistic filtering structure, the SRCKP achieves higher fidelity and robustness than either method alone. The inclusion of OSC and PLP further enhances practical applicability by addressing specific signal distortion and data transmission issues. This approach improves situational awareness for remote drivers, supporting the commercialization and reliability of automated vehicle systems that rely on remote intervention.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• teleoperation remote driving