Ephemeral Object Filtering in 3D Sensor Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle sensor data often includes points associated with ephemeral objects, which complicates tasks like map creation and localization, as these data points can render sensor data unusable or degrade accuracy, requiring manual and labor-intensive processes to remove them.
Innovation Solution
A method involving 2D image data and 3D sensor data processing to identify and filter out ephemeral objects by mapping data points to a grid of voxels, determining voxel statistics, and removing data points within voxels that meet specific threshold conditions, thereby isolating and removing ephemeral data points.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual processes are used to remove ephemeral object data points, then data accuracy for map creation is improved, but labor intensity and time consumption increase
Solution Approach 1:
The patent replaces manual mechanical processes with an automated computational system that uses sensor data fusion (LIDAR and camera data), voxel-based spatial partitioning, and machine learning classification to automatically identify and remove ephemeral object data points. This substitution eliminates human labor while maintaining high accuracy through algorithmic processing of spatial and temporal patterns in the sensor data.
Solution Approach 2:
The system performs self-service by automatically processing sensor data through the voxel grid methodology and machine learning models to identify and filter ephemeral objects without human intervention. The algorithm autonomously completes the entire workflow from data ingestion to filtered output generation, making the process self-sufficient and eliminating dependency on manual operations.
2Quantity of substance
If all sensor data points are retained for map creation, then data completeness is improved, but data usability deteriorates due to ephemeral objects
Solution Approach 1:
The patent segments the sensor data space into discrete voxels (three-dimensional grid cells) and further classifies each voxel as containing ephemeral objects, non-ephemeral objects, or empty space. This segmentation enables selective retention of data points based on their spatial classification, preserving complete information about permanent structures while excluding transient objects that would compromise map usability.
Solution Approach 2:
The system extracts and removes only the ephemeral object data points from the complete sensor dataset through the voxel-based filtering process. By taking out specifically the problematic ephemeral elements while retaining all other data, the system maintains data completeness for permanent features while eliminating usability issues caused by transient objects.
3Measurement precision
If complex filtering criteria are applied to remove ephemeral objects, then data accuracy is improved, but computational complexity increases
Solution Approach 1:
The voxel grid segmentation divides the complex three-dimensional space into manageable discrete units, allowing filtering criteria to be applied independently to each voxel. This breakdown transforms a potentially overwhelming global filtering problem into numerous simpler local decisions, improving accuracy through systematic evaluation while reducing computational complexity through problem decomposition.
Solution Approach 2:
The system performs preliminary actions by organizing sensor data into a voxel grid structure and pre-classifying voxels before the actual filtering operation. This preliminary organization and classification establish a structured framework that simplifies subsequent filtering operations, enabling accurate ephemeral object removal while maintaining computational efficiency through pre-processed spatial indexing.
Data Source
AI summary
Examples disclosed herein may involve (i) obtaining 2D image data and 3D sensor data that is representative of an area, (ii) identifying a first set of pixels associated with ephemeral objects detected in the area and a second set of pixels associated with non-ephemeral objects detected in the area, (iii) identifying a first set of ephemeral 3D data points associated with the detected ephemeral objects and a second set of non-ephemeral 3D data points associated with the detected non-ephemeral objects, (iv) mapping the first and second sets of 3D data points to a grid of voxels associated with the area, (v) making a determination that one or more voxels in the grid each contain a threshold extent of ephemeral data points, and (vi) based at least in part on the determination, filtering the 3D sensor data to remove the 3D data points contained within the one or more voxels.