LiDAR Point Cloud Compression for Dynamic Object Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional LiDAR data compression techniques fail to accurately represent dynamic objects and introduce inaccuracies due to over-representation removal and surface modeling, leading to difficulties in handling large volumes of data generated by advanced LiDAR sensors.
Innovation Solution
A system for live point cloud compression and decompression using a compression circuit that determines velocity for each cell in the LiDAR data, selectively compressing data based on velocity changes, and a decompression circuit that reconstructs data from compressed streams by extrapolating values using prior samples, ensuring accurate representation of dynamic objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional compression techniques remove over-represented data or model surfaces with partial functions, then data volume is reduced, but accuracy of dynamic object representation deteriorates
Solution Approach 1:
The patent applies dynamics by making the compression approach adaptive rather than static. The system dynamically determines whether to compress or retain each cell's data based on velocity calculations. Cells with high velocity (dynamic objects) are retained without compression, while cells with low velocity (stationary environment) are compressed using velocity-based prediction. This dynamic decision-making process resolves the contradiction by preserving accuracy for dynamic objects while achieving compression for static regions.
2Measurement precision
If velocity-based selective compression is applied to each cell, then accuracy of dynamic object representation is improved, but device complexity increases
Solution Approach 1:
The patent segments the point cloud data into discrete cells, with each cell independently processed. This segmentation allows the complex velocity-based compression logic to be applied locally to small, manageable units rather than to the entire large dataset. By dividing the problem into cell-level segments, the system achieves high accuracy for dynamic objects while keeping the computational complexity tractable through localized processing.
Solution Approach 2:
The patent implements partial action by applying full compression only to cells where it is beneficial (low-velocity stationary regions), while leaving high-velocity cells uncompressed. This partial application of compression logic optimizes the balance between accuracy and complexity by focusing computational effort only where necessary, avoiding the excessive complexity that would result from uniformly applying complex compression to all cells.
3Measurement precision
If all LiDAR data is retained without compression, then measurement accuracy is maintained, but data storage and transmission demands increase
Solution Approach 1:
The patent applies local quality by treating different spatial regions (cells) with different compression strategies based on their local characteristics. Cells containing dynamic objects (high velocity) maintain full data quality, while cells containing static environment (low velocity) undergo compression. This localized differentiation resolves the contradiction by preserving measurement precision only where it matters (dynamic objects) while reducing overall data volume through selective compression of less critical regions.
Data Source
AI summary
A system for point cloud data compression including a LiDAR sensor and a compression circuit. The compression circuit receives data for rays from the LiDAR sensor, wherein the data is arranged in cells which map to corresponding rays, and each cell includes a range value. The compression circuit constructs a frame for the received data and, for each cell, determines a velocity associated with the range value for the cell, determines whether to compress the frame with respect to the cell based upon the velocity; and selectively compresses the frame with respect to the cell accordingly. Also disclosed is a system including a device and a server which optionally are co-located. The device gathers data about an environment which is used to generate a point cloud.