Infrared Visual SLAM Navigation Under Changing Light Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional visual SLAM systems for mobile devices face performance degradation under low or changing light conditions and fail to maintain reliable navigation over a multi-year service life due to unbounded map file growth and high computational costs of incremental map updates.
Innovation Solution
The implementation of a visual positioning system using a camera configured to detect infrared wavelengths, combined with a flood infrared illumination source for localization and a structured infrared illumination source for obstacle detection, coordinated by a processor to create a consistent visual SLAM map across varying lighting conditions, allowing for autonomous navigation in unstructured environments without prior exploration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If visual SLAM systems use standard cameras for navigation, then the system is simple and low-cost, but the system fails under low or changing light conditions
Solution Approach 1:
The patent changes the operational parameter of the camera from visible light to infrared wavelength detection, and introduces infrared illumination sources to actively control the lighting conditions. This allows the system to operate reliably across diverse lighting conditions by transitioning to a different spectral domain where illumination can be controlled independently of environmental light.
Solution Approach 2:
The patent introduces infrared illumination sources as intermediary elements that actively provide lighting for the infrared camera. These illumination sources mediate between the camera and the environment, ensuring consistent illumination conditions regardless of ambient lighting, thereby enabling reliable navigation under diverse lighting conditions.
2Measurement precision
If visual SLAM systems operate continuously to maintain navigation accuracy, then positioning accuracy is maintained, but map file size grows unboundedly and computational cost increases
Solution Approach 1:
The patent implements a map management strategy where previously captured infrared images are discarded after being processed to create the visual SLAM map, rather than continuously accumulating new images. This approach maintains positioning accuracy through continuous visual feedback while preventing unbounded map file growth by selectively discarding redundant data.
Solution Approach 2:
The patent employs periodic action by capturing images at regular intervals during navigation and using these periodic measurements to continuously update and refine the visual SLAM map. This periodic image capture and processing maintains positioning accuracy without requiring continuous operation, thereby managing computational load and storage requirements.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables reliable and efficient navigation in diverse lighting conditions, extending the service life of mobile devices by minimizing the impact of changing light sources and reducing computational costs through coordinated infrared illumination and imaging.
Implementation Method 1
at least one camera configured to detect infrared wavelengths... at least one infrared illumination source... A flood infrared illumination source may be used to illuminate objects in an environment
Data Source
AI summary
A visual positioning system for mobile devices includes at least one infrared camera, at least one infrared illumination source, and a processor that coordinates operation of the at least one camera and illumination sources. A flood IR infrared illumination source illuminates environmental objects for localization of the mobile device during a first camera exposure window, and a structured IR illumination source illuminate environmental objects for detection and mapping of obstacles during a second camera exposure window. A visual SLAM map is constructed with images obtained from a first camera, with a single map being useable for positioning and navigation across a variety of environmental lighting conditions. A method for training a robot includes receiving operator input signals configured to start and stop recording of defined routes during the guiding of the mobile robot along paths between different pairs of desired robot destinations, such that routes for subsequent use by the robot are generated for paths imaged while recording is active.