Link-Based Sensor Data Realignment for Accurate Environment Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inaccuracies in sensor data and generated maps can cause robots and devices to incorrectly determine their location, leading to operational failures or safety risks, especially in complex environments like cities, where human detection of errors is implausible.
Innovation Solution
A system for correcting sensor data alignment by determining links between sensor data sets from different poses, perturbing poses to minimize errors, and generating maps using transformations, which can be computed locally or distributed across vehicles to reduce computational and temporal costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If sensor data is collected over a large region to generate comprehensive maps, then the coverage and completeness of the map is improved, but the computational complexity and time required for processing increases
Solution Approach 1:
The patent divides the large-scale map generation task into multiple overlapping local maps collected by different vehicles. Each vehicle generates a local map of its surveyed region, and these local maps are subsequently merged into a global map. This segmentation allows parallel processing of multiple regions simultaneously, reducing overall computation time while maintaining comprehensive coverage.
Solution Approach 2:
The patent performs preliminary alignment and transformation of sensor data into local maps before merging them into a global map. By pre-processing and organizing data into manageable local map units with established coordinate transformations, the system avoids the computational burden of processing all sensor data simultaneously, thereby reducing total map generation time.
2Measurement precision
If sensor data from multiple vehicles is merged to improve map accuracy, then the precision and reliability of the map is improved, but the computational resources required increase
Solution Approach 1:
The patent processes sensor data from multiple vehicles by first creating individual local maps for each vehicle's survey region. These local maps are then merged using coordinate transformations based on surveyed common features. This segmented approach allows computational resources to be distributed across multiple independent processing tasks rather than requiring centralized processing of all data simultaneously, reducing peak computational resource requirements.
Solution Approach 2:
The patent uses surveyed common features (landmarks, GPS coordinates) as reference points to create coordinate transformations between different vehicles' local maps. By copying and matching these reference features across multiple datasets, the system achieves accurate alignment and merging without requiring intensive computational comparison of entire datasets, thus improving accuracy while conserving computational resources.
3Measurement precision
If links are added or modified in the sensor data alignment to correct errors, then the map accuracy is improved, but the computational time for re-aligning sensor data increases
Solution Approach 1:
The patent modifies only the specific links or transformations that contain errors rather than re-aligning the entire sensor dataset. By identifying and correcting individual problematic links between poses or local maps, the system maintains overall alignment accuracy while minimizing computational time, as only localized portions of the data structure require reprocessing.
Data Source
AI summary
Generating a map associated with an environment may include collecting sensor data received from one or more vehicles and generating a set of links to align the sensor data. A mesh representation of the environment may be generated from the aligned sensor data. A system may determine a proposed link to add, a proposed link deletion, and/or a proposed link alteration, and receive a modification comprising instructions to add, delete, or modify a link. Responsive to receiving a modification, the system may re-align a window of sensor data associated with the modification. The modification and/or sensor data associated therewith may be collected as training data for a machine learning model, which may be trained to generate link modification proposals and/or determine sensor data that may be associated with a poor sensor data alignment.