Global Situation Awareness Graph Merging for Vehicle Occlusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Vehicles' cameras may be occluded or blocked, hindering situation awareness due to obstacles or other conditions, making it difficult for a single vehicle to ascertain its surroundings effectively.
Innovation Solution
A method and system that utilize multiple cameras from different vehicles or infrastructure to generate a global situation awareness graph by merging individual situation awareness graphs, using weights based on principal components analysis and Mahalanobis Distances, to provide a comprehensive view of surroundings, and predict object trajectories using dynamic bipartite and probabilistic graphs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single vehicle's camera is used for situation awareness, then the system complexity is low, but the detection coverage and reliability are reduced due to occlusions and blocked views
Solution Approach 1:
The patent combines camera data from multiple vehicles to create a merged situation awareness graph. Individual situation awareness graphs from different vehicles are integrated through graph merging operations, allowing the system to overcome occlusions and blocked views by aggregating observations from multiple perspectives. This merging approach improves detection reliability without requiring a single complex camera system on one vehicle.
Solution Approach 2:
The system enables vehicles to serve multiple functions: each vehicle acts as both a data collector (capturing its own situation awareness) and a data contributor (providing its graphs to other vehicles). This multi-functionality allows the network of vehicles to collectively achieve comprehensive situation awareness, where each vehicle's relatively simple camera system contributes to the overall reliable detection for the group.
2Loss of information
If multiple cameras from different vehicles are used to generate a global situation awareness graph, then the detection coverage and situation awareness are improved, but the data processing and graph merging complexity increase
Solution Approach 1:
The patent segments the overall situation awareness task into individual situation awareness graphs, with each vehicle generating its own separate graph based on its local camera observations. This segmentation allows each vehicle to process only its own data independently, reducing individual processing complexity. The segmented graphs are then merged to form the complete global situation awareness, maintaining information completeness while distributing the processing burden.
Solution Approach 2:
The situation awareness graph serves as an intermediary data structure that facilitates the integration of information from multiple vehicles. Rather than directly processing and combining raw camera data from all vehicles (which would be computationally intensive), the system uses intermediate graphs that summarize each vehicle's observations. These intermediate representations are then merged through graph operations, reducing the overall data processing complexity while preserving essential information.
3Measurement precision
If weights are assigned to different situation awareness graphs based on principal components analysis and Mahalanobis distances, then the merging accuracy and object detection precision are improved, but the computational requirements increase
Solution Approach 1:
The patent changes the parameter space by transforming raw camera observations into situation awareness graphs with specific mathematical properties (using principal components analysis and Mahalanobis distances). This parameter transformation enables more accurate weighting and merging of graphs from different vehicles. By operating in this transformed parameter space rather than raw pixel data, the system achieves higher detection precision while managing computational requirements through efficient graph-based representations.
Data Source
AI summary
In various embodiments, methods, systems, and vehicles are provided that include obtaining first camera images from a first camera onboard a first vehicle; generating, via one or more computer processors, a first situation awareness graph with respect to objects near the first vehicle, using the first camera images; obtaining second camera images from a second camera of a second device that is in proximity to the first vehicle; generating, via one or more computer processors, a second situation awareness graph with the respect to the objects, using the second camera images; and generating, via one or more computer processor, a global situation awareness graph with respect to the objects, by merging the first situation awareness graph with the second situation awareness graph, using respective first and second weights for the first and second situation awareness graphs.