Motion recovery in light-attenuating media from image-shading variations

Area of Science:

• Robotics
• Computer Vision
• Optical Sensing

Background:

• Accurate vehicle positioning and motion estimation are crucial for robotic autonomy.
• Passive-vision sensing offers a method for extracting environmental and motion data from images.
• Underwater operations present unique challenges due to artificial lighting and light attenuation.

Purpose of the Study:

• To investigate motion recovery from image-shading variations for robotic vehicles in challenging environments.
• To develop a model for image brightness that accounts for illumination attenuation and variation.
• To determine vehicle motion using optical beacons in a specific undersea application configuration.

Main Methods:

• Utilizing an established image model incorporating illumination effects.
• Analyzing image-shading variations from Lambertian planar surfaces.
• Developing a closed-form solution for motion determination.

Main Results:

• Demonstrated closed-form determination of vehicle motion.
• Successfully recovered motion using three Lambertian planar surfaces as optical beacons.
• Validated the effectiveness of the image model for motion recovery under simulated undersea conditions.

Conclusions:

• The proposed method enables accurate motion recovery for robotic vehicles in environments with artificial lighting and attenuation.
• Passive-vision sensing, combined with image-shading analysis, provides a viable solution for undersea navigation.
• The use of three Lambertian planar surfaces as optical beacons simplifies and enhances motion determination.