Model-Based Scan Line Encoder for LIDAR Data Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing LIDAR scanning technologies face inefficiencies due to mechanical misalignments and offsets, leading to large deviations in scan data conversion and inefficiencies in naive encoding techniques, resulting in artifacts like voids in scan lines.

Innovation Solution

A model-based scan line encoding method is introduced, which defines a geometry model for describing scan lines, calculates a trajectory model to represent deviations, and compresses residuals to enable effective lossless and lossy compression, matching data loss with native error sources, and allows quick location of scan data without uncompressing entire datasets.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If naive encoding techniques are used to store scan data, then the implementation is simple, but the storage space required is large and artifacts like voids appear in scan lines

Engineering Contradiction:
Improveencoding simplicityVSAvoidstorage space
Core Design Contradiction:
Ease of manufactureVSQuantity of substance

Solution Approach 1:

The scan data is segmented into scan lines, and each scan line is further segmented into groups of points. A separate encoding model is applied to each scan line, allowing independent optimization and reducing the overall storage requirements while maintaining simplicity in implementation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms the scan data from raw coordinate format into a parametric representation using polynomial models. By changing the parameters from storing all point coordinates to storing model coefficients and deviation values, the storage space is significantly reduced while eliminating artifacts like voids.

Inventive Principle:
Principle #35Parameter changes

2Area of stationary object

If mechanical positioning is used to scan the LIDAR, then the system can cover a large area, but misalignments and offsets cause large deviations in scan data

Engineering Contradiction:
Improvescan coverage areaVSAvoidscan data accuracy
Core Design Contradiction:
Area of stationary objectVSMeasurement precision

Solution Approach 1:

The patent incorporates feedback by calculating the deviation of actual scan points from the polynomial trajectory model and storing these residuals. This feedback mechanism allows the system to compensate for mechanical misalignments and offsets by correcting the idealized model with measured deviations, thereby improving scan data accuracy while maintaining large coverage area.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The patent performs preliminary action by establishing a polynomial trajectory model that represents the ideal scan path before processing individual scan points. This pre-established model serves as a reference framework that accounts for expected mechanical behavior, allowing subsequent deviations to be measured and corrected systematically.

Inventive Principle:
Principle #10Preliminary action

3Loss of information

If entire datasets are uncompressed for access, then all data is available, but the access time is slow and storage space is wasted

Engineering Contradiction:
Improvedata availabilityVSAvoiddata access time
Core Design Contradiction:
Loss of informationVSLoss of time

Solution Approach 1:

The patent segments the compressed data structure into scan line-level units that can be independently accessed and decompressed. This allows the system to load and process only the specific scan lines needed for a particular query, rather than uncompressing the entire dataset, thereby reducing access time and preserving data availability for selective access.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2972472B1Model based scan line encoder
Publication Date: 2021.04.28 LEICA GEOSYSTEMS AG
  • EP2972472B1 patent drawingFigure 1
  • EP2972472B1 patent drawingFigure 2
  • EP2972472B1 patent drawingFigure 3

AI summary

A model-based scan line encoder is disclosed. A method of model-based scan line encoding includes defining a geometry model for describing a scan line of a scan, the scan line including multiple scan points. The method further includes calculating a trajectory model representing an approximate pattern of deviation of the multiple scan points relative to the geometry model. The method further includes calculating multiple residuals, each of the residuals associated with a difference between the deviation of the scan points and the trajectory model. The method may further include compressing the residuals.