Multi-Journey 3D Mapping for Calibration-Resilient Vehicle Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to create precise 3D maps for autonomous vehicles using crowdsourced data due to dynamic changes in road conditions and the need for expensive, calibrated equipment.
Innovation Solution
A method and system for building precise 3D maps using multi-journey data from multiple vehicles, incorporating multi-journey association and refinement to optimize 6 degrees of freedom landmark poses and camera parameters, utilizing spectral and hierarchical clustering to improve accuracy and resilience to calibration errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If crowdsourced data from multiple vehicles is used to create 3D maps, then map precision and adaptability are improved, but data complexity and processing difficulty increase
Solution Approach 1:
The patent segments the complex multi-vehicle mapping problem into distinct processing stages: individual vehicle journey processing, data aggregation from multiple sources, and iterative refinement cycles. Each stage handles specific aspects of the data independently, making the overall complex system manageable and processable through modular operations
Solution Approach 2:
The patent implements dynamic iterative refinement where map data is continuously updated and improved through multiple processing passes. The system dynamically adjusts by reprocessing data with refined parameters, allowing the mapping system to adapt and improve precision through repeated optimization cycles rather than static single-pass processing
2Reliability
If multi-journey data from multiple vehicles is aggregated, then map accuracy and resilience to calibration errors are improved, but computational requirements and processing time increase
Solution Approach 1:
The patent performs preliminary processing of individual vehicle journeys independently before aggregation, pre-computing and preparing data structures that can be efficiently combined. This preliminary organization of multi-vehicle data reduces the computational burden during the final integration and refinement stages, decreasing overall processing time
Solution Approach 2:
The patent implements continuous iterative refinement where processing continues across multiple passes without complete re-computation. Each iteration builds upon previous results, maintaining and improving accuracy through continuous optimization rather than discrete restarts, thereby reducing total processing time while enhancing map reliability
3Ease of manufacture
If expensive calibrated equipment is replaced with crowdsourced vehicle data, then system cost is reduced, but measurement precision and data quality may deteriorate
Solution Approach 1:
The patent merges data from multiple independent vehicle sources to compensate for individual data quality limitations. By combining numerous lower-precision measurements from crowd-sourced vehicles through aggregation and iterative refinement, the system achieves map precision comparable to expensive calibrated equipment while dramatically reducing system costs
Solution Approach 2:
The patent implements feedback mechanisms where processed map data is continuously compared and refined against incoming multi-vehicle observations. This feedback loop identifies and corrects errors in crowd-sourced data, maintaining high measurement precision despite the use of lower-cost sensing equipment by constantly validating and adjusting results
Data Source
AI summary
A method performed by an apparatus is described. The method includes receiving map data that is based on first image data, second image data, and a similarity metric. The first image data can be received from a first vehicle and represent an object. The second image data can be received from a second vehicle and represent the object. The similarity metric can be associated with the object represented in the first image data and the object represented in the second image data. The method can also include storing, by a vehicle, the received map data and localizing the vehicle based on the stored map data.