Neural Network Object Selection for V2X Bandwidth Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current vehicle-to-everything (V2X) technologies face challenges in efficiently communicating and processing the vast amount of data required for autonomous driving and advanced driver-assistance systems, leading to bandwidth limitations and unnecessary data transmission.
Innovation Solution
A computer-implemented method using a trained neural network to select and filter relevant objects from sensor data, reducing the number of objects communicated between vehicles by identifying only those that trigger corrective actions in autonomous driving systems, thereby optimizing data exchange and processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all vehicles broadcast all detected objects to all other vehicles, then complete information is shared, but bandwidth requirements become prohibitively large
Solution Approach 1:
The patent extracts and transmits only the most relevant information (critical objects and events) rather than all detected data. The system identifies and filters out redundant information, sending only essential data about objects that pose potential risks or require attention, thereby reducing bandwidth consumption while maintaining information completeness for safety-critical decisions.
Solution Approach 2:
The patent applies different transmission priorities and qualities to different types of data. Critical safety-related objects receive higher priority and more detailed transmission, while less important objects are either summarized or omitted. This local differentiation of data quality allows the system to optimize bandwidth usage based on the actual importance of each data element.
2Quantity of substance
If 5G networks provide higher bandwidth, then more data can be transmitted, but processing capacity remains limited
Solution Approach 1:
The patent performs preliminary filtering, aggregation, and prioritization of data at the source vehicle before transmission. By pre-processing the data and identifying critical information in advance, the system reduces the processing burden on receiving vehicles and network infrastructure, enabling real-time response despite limited processing capacity at any single node.
Solution Approach 2:
The patent introduces an intermediary processing layer that aggregates data from multiple sources, filters redundant information, and prioritizes critical events before distribution. This intermediary function distributes the processing load across the network rather than requiring every vehicle to process all incoming data, thereby improving overall system productivity.
3Length of stationary object
If sensor detection range is extended to ±80m, then more objects are detected, but the number of objects to communicate increases significantly
Solution Approach 1:
The patent extracts only the essential attributes of detected objects (position, velocity, acceleration, type) and transmits only those objects that meet specific relevance criteria. By filtering out redundant object details and focusing on critical motion parameters, the system reduces the effective number of objects that need full communication while maintaining extended detection capabilities.
Solution Approach 2:
The patent detects all objects within extended range but communicates only a partial subset that is deemed relevant based on motion characteristics and spatial relationships. This partial communication approach allows the system to maintain comprehensive situational awareness through extended sensing while reducing communication overhead by transmitting only the necessary portion of detected data.
Data Source
AI summary
Disclosed is a computer-implemented method of training a neural network to select objects in a vicinity of a target vehicle, including: aggregating sensor data related to the plurality of vehicles; filtering the sensor data according to one or more conditions, said conditions identifying an action of an ADAS/AD system of at least one of the plurality of vehicles; identifying one or more objects in the vicinity of target vehicle based on the filtered sensor data; and using the identified one or more objects to train the neural network to determine potential objects that cause a triggering of an ADAS/AD system of another vehicle.