Vehicle Image Segmentation for Sparse-Map Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in navigating due to the vast amounts of data they need to process and store, particularly with traditional mapping technologies, which can limit their ability to efficiently analyze and update maps while traveling.
Innovation Solution
The use of cameras and processors to analyze images and determine vehicle positions, boundaries, and distances, generating navigational responses based on pixel analysis, and employing sparse maps that require less data storage and transfer, allowing for efficient navigation without extensive data storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional mapping technology is used to navigate, then navigation capability is provided, but the volume of data needed to store and update the map becomes extremely large
Solution Approach 1:
The patent segments the map data into two types: sparse map data (stored permanently) and dense map data (temporarily cached). The sparse map contains only essential navigation information that is periodically updated, while dense map data is generated on-demand and discarded after use. This segmentation reduces the permanent storage requirement from terabytes to megabytes while maintaining navigation capability.
Solution Approach 2:
The patent changes the update frequency parameter of map data from continuous/traditional updates to periodic updates (e.g., every 100 miles or hourly). This parameter change allows the system to navigate effectively with less frequent map updates, significantly reducing the data volume that needs to be stored and transferred while maintaining adequate navigation accuracy.
2Measurement precision
If vast volumes of data are collected and analyzed, then navigation accuracy is improved, but processing time and computational burden increase
Solution Approach 1:
The patent applies partial action by processing only the necessary portion of data for current navigation needs. Instead of analyzing all collected sensor data and map information, the system selectively processes only the sparse map data and relevant sensor inputs required for immediate navigation decisions, reducing processing time while maintaining adequate accuracy.
Solution Approach 2:
The patent performs preliminary action by pre-processing and compressing map data into sparse representations before they are needed for navigation. The sparse map is created in advance with only essential navigation features, so when navigation is required, the system does not need to process large volumes of raw map data, significantly reducing real-time processing time.
3Reliability
If traditional map updates are performed frequently, then map accuracy is maintained, but data transfer and storage requirements increase
Solution Approach 1:
The patent segments map updates into periodic sparse updates rather than continuous dense updates. The sparse map contains only critical navigation elements that are updated at extended intervals (e.g., every 100 miles or hourly), reducing data transfer volume from gigabytes to kilobytes per update while maintaining sufficient map accuracy for navigation.
Solution Approach 2:
The patent changes the update interval parameter from frequent/traditional updates to extended periodic updates. By increasing the time and distance between map updates and reducing the amount of data transferred per update, the system maintains adequate map accuracy while dramatically reducing data transfer and storage requirements.
Data Source
AI summary
Systems and methods for navigating a host vehicle are disclosed. In one implementation at least one processor is programmed to receive two or more images captured by a camera of the host vehicle from an environment of the host vehicle; analyze the two or more images to identify a representation of at least a portion of a first object and a representation of at least a portion of a second object; determine a first region of the two or more images associated with the first object and a type of the first object; and determine a second region of the two or more images associated with the second object and type of the second object, wherein the type of the first object is different from the type of the second object.