Teleoperated Driving: a New Challenge for 3D Object Detection in Compressed Point Clouds

2025-06-13

AI Summary

This work addresses the safety-critical need for real-time, robust 3D vehicle and pedestrian detection from compressed point clouds in bandwidth-constrained V2X-based teleoperation (TD) systems. Method: We propose an end-to-end evaluation framework aligned with 3GPP TD requirements (end-to-end latency <100 ms, rate constraints), integrating state-of-the-art point cloud compression techniques (PCL, Octree, LZ4) with leading 3D detectors (PointPillars, SECOND). We systematically benchmark their joint performance and introduce SELMA-Ext—a newly constructed, publicly released dataset featuring fine-grained 3D annotations. Contribution/Results: Our cross-layer co-optimization approach establishes the first TD-specific open benchmark and evaluation methodology. Quantitative analysis reveals the compression–accuracy trade-off frontier: achieving >92% point cloud compression while maintaining mAP >65%, with end-to-end latency fully compliant with 3GPP standards.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Learning on the Edge & Model Compression
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

In recent years, the development of interconnected devices has expanded in many fields, from infotainment to education and industrial applications. This trend has been accelerated by the increased number of sensors and accessibility to powerful hardware and software. One area that significantly benefits from these advancements is Teleoperated Driving (TD). In this scenario, a controller drives safely a vehicle from remote leveraging sensors data generated onboard the vehicle, and exchanged via Vehicle-to-Everything (V2X) communications. In this work, we tackle the problem of detecting the presence of cars and pedestrians from point cloud data to enable safe TD operations. More specifically, we exploit the SELMA dataset, a multimodal, open-source, synthetic dataset for autonomous driving, that we expanded by including the ground-truth bounding boxes of 3D objects to support object detection. We analyze the performance of state-of-the-art compression algorithms and object detectors under several metrics, including compression efficiency, (de)compression and inference time, and detection accuracy. Moreover, we measure the impact of compression and detection on the V2X network in terms of data rate and latency with respect to 3GPP requirements for TD applications.

Problem

• Detecting cars and pedestrians in compressed point clouds for Teleoperated Driving safety
• Evaluating compression algorithms and object detectors for efficiency and accuracy
• Assessing V2X network impact on data rate and latency for TD

Innovation

• Utilizes SELMA dataset for 3D object detection
• Evaluates compression algorithms and object detectors
• Measures V2X network impact on teleoperated driving