Risk Occupancy: A New and Efficient Paradigm through Vehicle-Road-Cloud Collaboration

TL;DR

The paper proposes 4D Risk Occupancy, a risk-aware occupancy representation that fuses road surface geometry, dynamic/static risk factors, and temporal prediction into a BEV-compatible map for intelligent connected vehicles. Grounded in a Vehicle-Road-Cloud framework, it quantifies risk via ETA-based dynamics, samples the road with a fine grid, and propagates risk into a local path-planning strategy that favors collision-free dynamic node sets and Gaussian-smoothed trajectories. The approach is validated on the DAIR-V2X dataset, showing qualitative robustness across right-turn, straight, and left-turn scenarios and quantitative gains in safety redundancy (+12.5%) and reduced braking deceleration (-5.41%) at an initial speed of 8 m/s. The cloud-assisted architecture enables beyond-line-of-sight perception and efficient, real-time planning by offloading heavy computation while maintaining low latency at the vehicle end. Overall, the work advances a unified perceptual paradigm and a practical planning pipeline for safer and more efficient autonomous driving in mixed traffic environments.

Abstract

This study introduces the 4D Risk Occupancy within a vehicle-road-cloud architecture, integrating the road surface spatial, risk, and temporal dimensions, and endowing the algorithm with beyond-line-of-sight, all-angles, and efficient abilities. The algorithm simplifies risk modeling by focusing on directly observable information and key factors, drawing on the concept of Occupancy Grid Maps (OGM), and incorporating temporal prediction to effectively map current and future risk occupancy. Compared to conventional driving risk fields and grid occupancy maps, this algorithm can map global risks more efficiently, simply, and reliably. It can integrate future risk information, adapting to dynamic traffic environments. The 4D Risk Occupancy also unifies the expression of BEV detection and lane line detection results, enhancing the intuitiveness and unity of environmental perception. Using DAIR-V2X data, this paper validates the 4D Risk Occupancy algorithm and develops a local path planning model based on it. Qualitative experiments under various road conditions demonstrate the practicality and robustness of this local path planning model. Quantitative analysis shows that the path planning based on risk occupation significantly improves trajectory planning performance, increasing safety redundancy by 12.5% and reducing average deceleration by 5.41% at an initial braking speed of 8 m/s, thereby improving safety and comfort. This work provides a new global perception method and local path planning method through Vehicle-Road-Cloud architecture, offering a new perceptual paradigm for achieving safer and more efficient autonomous driving.

Figures (9)

• Figure 1: When the green light comes on and a pedestrian runs across the road, the white ICV fails to recognize the runner due to the truck on the right front. Vehicle-Road-Cloud collaboration enables ICV to access the risk occupancy conditions for the next about 3 seconds, perceiving potential risks in advance.
• Figure 2: The curvilinear relationship between risk value and ETA.
• Figure 3: Demonstration of the road sampling point layout. The sampling points on the road are arranged along the direction of the road with an adjustable resolution, and different sizes of resolution can be selected based on traffic flow density and computational resources.
• Figure 4: Three images elucidate the 4D Risk Occupancy outcomes derived from the DAIR-V2X dataset. The image on the left depicts the perception result in BEV, providing an original perspective. The middle image presents the 4D Risk Occupancy map in BEV, offering an intuitive analysis of risk distribution. The image on the right transitions to a three-dimensional view, showcasing the 4D Risk Occupancy map in a format that enhances spatial understanding.
• Figure 5: The three figures show the cloud platform's pre-set potential node sets for an ICV to navigate intersections, corresponding to three driving scenarios: right turn, straight, and left turn.