Hazardous Road Object Sharing in HD Maps With Octree Resolution Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous vehicle systems waste network bandwidth by transmitting differential features corresponding to small, non-hazardous road objects, and there is a need for efficient detection and communication of hazardous road objects in high-definition maps.
Innovation Solution
Utilizing octree representations to compress and transmit low-resolution volumetric deviations of hazardous objects via vehicle-to-vehicle links, and high-resolution data to a cloud computing system for integration into updated maps, with size and temporal existence-based classification and consensus building among vehicles and servers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all differential features are transmitted to the server for HD map updates, then map accuracy is improved, but network bandwidth is wasted on small non-hazardous objects
Solution Approach 1:
The patent segments the transmission process into two paths: vehicle-to-vehicle transmission for immediate hazard response and vehicle-to-infrastructure transmission for map updates. By segmenting the data flow, the system transmits only essential hazard information to nearby vehicles while filtering out non-hazardous objects before server transmission, thus reducing unnecessary bandwidth consumption while maintaining map accuracy for hazardous objects.
Solution Approach 2:
The patent extracts and separates hazardous objects from non-hazardous differential features using classification algorithms. Only the extracted hazardous objects are prioritized for transmission and map updates, while non-hazardous objects are filtered out. This extraction process enables the system to maintain map accuracy for critical hazards while reducing overall network bandwidth consumption by excluding irrelevant data.
2Measurement precision
If high-resolution volumetric data is transmitted for all detected objects, then detection accuracy is improved, but communication time and bandwidth increase
Solution Approach 1:
The patent applies local quality by transmitting high-resolution volumetric data only for hazardous objects and low-resolution or compressed data for non-hazardous objects. The system dynamically adjusts data resolution based on object hazard classification, ensuring detection accuracy for critical hazards while reducing communication time and bandwidth for benign objects through selective resolution transmission.
Solution Approach 2:
The patent implements dynamic data transmission by adjusting the resolution and amount of volumetric data sent based on real-time hazard assessment. The system dynamically switches between high-resolution transmission for hazardous objects and compressed/low-resolution transmission for non-hazardous objects, optimizing the balance between detection accuracy and communication efficiency under varying conditions.
3Reliability
If real-time volumetric map updates are performed for all vehicles, then safety is improved, but network load increases
Solution Approach 1:
The patent introduces nearby vehicles as intermediaries between the detecting vehicle and the server. Instead of all vehicles receiving full updates from the server, nearby vehicles first receive and process hazard information locally through vehicle-to-vehicle communication. This intermediary approach enables real-time safety updates for hazardous objects while reducing overall network load by distributing the update burden across multiple vehicles rather than centralizing it.
Solution Approach 2:
The patent performs preliminary hazard detection and classification locally at each vehicle before server communication. By pre-processing and identifying hazardous objects locally, the system prepares only essential hazard information for transmission to the server and nearby vehicles, rather than transmitting complete volumetric data for all detected objects. This preliminary action reduces network load while maintaining real-time safety through local processing.
Data Source
Figure 1A~1D
Figure 2~3
Figure 4~5
AI summary
Systems, apparatuses and methods may provide for vehicle technology that detects one or more differences between a crowdsourced map of an ambient environment and a real-time volumetric map of the ambient environment and sends a first message via a vehicle-to-vehicle (V2V) link, wherein the difference(s) are represented in the first message at a first resolution. Additionally, the vehicle technology sends a second message via a vehicle-to-infrastructure (V2I) link, wherein the difference(s) are represented in the second message at a second resolution, and wherein the first resolution is less than the second resolution. Moreover, server technology may integrate a first octree representation and a second octree representation into a dynamic layer associated with the crowdsourced map.