Sensor Data Alignment for Accurate Environment Mapping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inaccuracies in sensor data and maps generated by robots and autonomous devices can lead to incorrect location determination, potentially causing operational issues or safety risks, especially in complex environments like cities, where human detection of errors is impractical due to the complexity and volume of data.
Innovation Solution
A system that includes a localization component, a mapping component, and a correction component to align sensor data from different poses, allowing for the addition, deletion, or adjustment of links between sensor data points, using machine-learning models to identify and correct errors, thereby improving map accuracy and reducing computational resources.
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 complexity and volume of data increases making error detection impossible for humans
Solution Approach 1:
The patent introduces automated computational systems and algorithms as intermediaries between the complex sensor data and human operators. These systems automatically detect errors, align sensor data from multiple poses, and generate corrections, replacing the need for human error detection in large-scale maps while maintaining map coverage.
Solution Approach 2:
The patent replaces manual human review and error detection processes with automated computational mechanisms. Machine learning models and alignment algorithms automatically process sensor data, identify inconsistencies, and correct errors without human intervention, enabling handling of large-scale data that exceeds human capacity.
2Ease of operation
If traditional methods are used to align sensor data and generate maps, then the process is simple to understand, but the time to generate accurate maps is excessive and computational resources are intensive
Solution Approach 1:
The patent transforms the sensor data alignment problem by changing the approach from traditional point-by-point manual alignment to pose-based batch alignment. By organizing data around vehicle poses and using automated alignment algorithms, the system achieves both simplicity and speed, reducing map generation time while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary organization of sensor data by poses and trajectories before alignment. By pre-structuring the data and identifying candidate alignments in advance, the system reduces the computational burden during the actual alignment process, decreasing overall map generation time while maintaining operational simplicity.
3Reliability
If comprehensive sensor data alignment is performed to ensure map accuracy, then the reliability of location determination is improved, but the computational power required becomes excessive
Solution Approach 1:
The patent segments the sensor data alignment process into distinct stages: data collection by pose, candidate alignment identification, alignment computation, and error correction. This segmentation allows the system to process data in manageable chunks rather than attempting comprehensive simultaneous alignment, reducing peak computational power requirements while maintaining location determination accuracy.
Solution Approach 2:
The patent implements a multi-pass alignment approach where initial alignments are performed with standard precision, followed by targeted refinement of specific regions or poses that show higher uncertainty. This partial excessive action ensures high reliability where needed while avoiding unnecessary computational expenditure in well-aligned regions, optimizing the balance between accuracy and power consumption.
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.