AirV2X: Unified Air-Ground Vehicle-to-Everything Collaboration
This addresses the problem of high costs and coverage gaps in V2X systems for autonomous driving, though it is incremental as it focuses on dataset creation rather than a new method.
The authors tackled the limitations of traditional infrastructure-based V2X systems by introducing AirV2X-Perception, a large-scale dataset using UAVs to enhance multi-vehicular collaborative driving, resulting in a dataset of 6.73 hours of drone-assisted scenarios across diverse environments.
While multi-vehicular collaborative driving demonstrates clear advantages over single-vehicle autonomy, traditional infrastructure-based V2X systems remain constrained by substantial deployment costs and the creation of "uncovered danger zones" in rural and suburban areas. We present AirV2X-Perception, a large-scale dataset that leverages Unmanned Aerial Vehicles (UAVs) as a flexible alternative or complement to fixed Road-Side Units (RSUs). Drones offer unique advantages over ground-based perception: complementary bird's-eye-views that reduce occlusions, dynamic positioning capabilities that enable hovering, patrolling, and escorting navigation rules, and significantly lower deployment costs compared to fixed infrastructure. Our dataset comprises 6.73 hours of drone-assisted driving scenarios across urban, suburban, and rural environments with varied weather and lighting conditions. The AirV2X-Perception dataset facilitates the development and standardized evaluation of Vehicle-to-Drone (V2D) algorithms, addressing a critical gap in the rapidly expanding field of aerial-assisted autonomous driving systems. The dataset and development kits are open-sourced at https://github.com/taco-group/AirV2X-Perception.