Autonomous Vehicle Camera Data Normalization

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Standard camera data generated for human viewing is not suitable for input into machine learning models used in autonomous vehicles, requiring additional processing to normalize variances in sensor observations.

Innovation Solution

The implementation of camera data normalization techniques, including color normalization, spherical reprojection, and stabilization based on a registration point, to prepare camera data for machine learning models in autonomous vehicles.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If camera data is processed with color normalization, spherical reprojection, and stabilization, then machine learning model performance is improved, but processing complexity increases

Engineering Contradiction:

Improvemachine learning model performanceVSAvoidprocessing complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies color normalization, spherical reprojection, and stabilization as preliminary processing steps to camera data before it is fed into machine learning models. This preliminary action prepares the data in advance, ensuring consistent input quality and reducing the burden on the learning algorithms, thereby improving model performance while managing processing complexity through structured preprocessing.

Inventive Principle:

Principle #10Preliminary action

2Ease of manufacture

If standard camera lenses are used, then device cost is reduced, but measurement precision of object distance deteriorates

Engineering Contradiction:

Improvedevice costVSAvoidobject distance determination

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent replaces the need for specialized camera lenses or stabilization devices with computational methods. By applying spherical reprojection and stabilization algorithms to the image data, the system achieves accurate object distance determination and stable images without requiring expensive hardware modifications, thus maintaining low device cost while improving measurement precision.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If camera data is normalized for machine learning, then adaptability of the system is improved, but processing time increases

Engineering Contradiction:

Improvemachine learning compatibilityVSAvoidprocessing time

Core Design Contradiction:

Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent transforms camera data by changing key parameters including color space normalization, spherical coordinate reprojection, and stabilization transformations. These parameter changes make the data adaptable to machine learning models by presenting it in a standardized format, while the efficiency of the transformation algorithms helps manage the processing time overhead.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11704895B2Camera data normalization for an autonomous vehicle

Publication Date: 2023.07.18 APPLIED INTUITION INC

AI summary

Camera data normalization for an autonomous vehicle are described herein, including: receiving, from one or more cameras of the autonomous vehicle, camera data; applying a color normalization to the camera data; applying a spherical reprojection to the camera data; and applying, based on a registration point, a stabilization to the camera data.