Edge-Managed Object Grids for Timely Autonomous Vehicle Sharing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in sharing accurate and up-to-date object maps due to delays in wireless communication, format inconsistencies, and legal liabilities associated with stale or inaccurate data, which affects their navigation and safety.
Innovation Solution
A system where edge nodes in geographic zones manage and update an Autonomous System Grid (ASG) by filtering scene descriptions based on timestamps and accuracy, combining sensor data from vehicles and fixed infrastructure, and broadcasting a standardized, compressed, and encrypted ASG to ensure compatibility and safety across different fleets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If object maps are shared between autonomous vehicles using wireless communication systems, then information exchange and collaborative navigation are improved, but data staleness and accuracy deterioration occur due to communication delays
Solution Approach 1:
The system performs preliminary classification and filtering of objects at the source vehicle before transmission. Objects are pre-processed, categorized by importance, and formatted according to a common schema in advance, so that when data is transmitted over the delayed wireless channel, it is already optimized for immediate use and less susceptible to staleness issues.
Solution Approach 2:
A standardized common object map format acts as an intermediary layer between different vehicle perception systems. This intermediate representation layer translates diverse sensor data into a unified format, enabling reliable information exchange while abstracting away the timing and source variations that would otherwise compromise accuracy.
2Adaptability or versatility
If object maps from different autonomous vehicles are combined, then comprehensive environmental awareness is improved, but format inconsistencies and data integration challenges arise
Solution Approach 1:
The patent establishes a universal common object map format that can represent objects detected by any sensor type (LIDAR, camera, radar) and any classification algorithm. This single standardized format serves multiple functions: it accommodates diverse input sources, enables seamless data fusion, and simplifies downstream processing across different vehicle platforms.
Solution Approach 2:
The system transforms heterogeneous object map data from various vehicles into a standardized parameter set with consistent fields, data types, and schemas. By changing the parameter representation to a common format, the system enables straightforward data integration while filtering out format-related complexities.
3Reliability
If standardized object map formats are implemented across different fleets, then interoperability and legal liability clarity are improved, but implementation complexity and transition costs increase
Solution Approach 1:
The standardization implementation is segmented into modular components: a core common format definition, optional extension schemas for fleet-specific requirements, and progressive adoption phases. This segmentation allows different fleets to implement the standard at their own pace while maintaining interoperability, reducing the complexity burden on individual organizations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Embodiments are disclosed for sharing classified objects perceived by autonomous vehicles. In an embodiment, a method comprises: obtaining, from an autonomous vehicle (AV) 1305, 1308, a first scene description, the first scene description 1312 including one or more classified objects 1309, 1310, 1311 detected in a first zone (Zone 2) of a geographic area (e.g., a city); updating, by a first edge node 1303 in the first zone and using a plurality of scene descriptions, an autonomous system grid (ASG) 1316 for the first zone, the ASG being stored in storage device 1315; and sending, by the first edge node 1303, the updated ASG to a second edge node 1302 located in the first zone or in a second zone (Zone 1) of the geographic area. Edge nodes 1302, 1303 can be AVs or other vehicles. Network 1300 includes mobile services platform (MSP) 1301 coupled to edge nodes 1302, 1303 through core network 1304. In an embodiment, all edge nodes in network 1300 communicate with each either directly through a wired or wireless communications link, or through MSP 1301. For example, AV 1305 traveling on road segment 1306 can receive ASG update 1317 directly from edge node 1303 and/or MSP 1301 through a high-speed wired backhaul. Using the planned route, the edge node can select and send to the AV the ASGs for all the zones which contain at least a portion of the planned route for the AV. Using the ASGs from the edge node, the AV can plan a new route to avoid delays through the geographic area. Additionally, having the ASGs stored in the AV allows for delta updates from each edge node along the planned route when the AV is within communication range of the edge node. A delta update includes only the changed portions of the ASG stored at the edge node rather than the entire ASG. This allows for faster data transport. Updating ASG 1316 includes but is not limited to: adding objects, deleting objects, updating the location of moving objects, updating the speed, velocity, acceleration or heading of the objects and updating labels for the objects to more accurately describe the objects. ASG 1316 is also timestamped before it is broadcast from edge node 1303. In an embodiment, ASG 1316 is assigned a confidence score indicating its accuracy.