Robot Keyframe Matching Using Pixel Descriptors for Pose Changes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robotic vision systems rely on feature-based approaches for object detection, which are limited in accuracy when images lack distinctive features, and struggle to perform tasks when the starting point or orientation of objects does not align with programmed tasks.
Innovation Solution
The system captures images and identifies keyframes, comparing pixel descriptors rather than features to improve matching accuracy, allowing robots to perform tasks even when the environment or object orientation changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If feature-based approaches are used for object detection, then the system is simpler to implement, but accuracy deteriorates when images lack distinctive features
Solution Approach 1:
The patent changes the fundamental parameter of comparison from feature-based to pixel-based matching. Instead of extracting and comparing discrete features, the system compares pixel descriptors directly, transforming the detection approach to achieve higher accuracy without proportionally increasing implementation complexity
Solution Approach 2:
The patent transitions from comparing a limited set of extracted features to comparing comprehensive pixel-level descriptors across the entire image. This dimensional expansion from feature space to pixel space provides much richer information for accurate matching, especially in images lacking distinctive features
2Use of energy by moving object
If feature-based approaches are used, then the system requires less computational resources, but reliability deteriorates when object orientation or starting point changes
Solution Approach 1:
The patent creates a universal matching approach that works regardless of object orientation, position, or scale. By comparing pixel descriptors directly rather than relying on pre-defined features, the system achieves orientation-invariant and position-invariant matching, making it universally applicable to various task configurations
Solution Approach 2:
The system performs preliminary keyframe selection and pixel descriptor computation to create a robust reference library. By pre-processing and storing pixel-level information from keyframes, the system enables reliable real-time matching without requiring excessive computational resources during task execution
3Measurement precision
If pixel-level comparison is implemented, then accuracy of object detection improves, but device complexity increases
Solution Approach 1:
The patent segments the complex pixel-level comparison task into manageable components: keyframe selection, pixel descriptor extraction, descriptor comparison, and task execution. This segmentation allows the system to achieve high accuracy through structured processing without overwhelming system complexity
Solution Approach 2:
The system creates simplified copies of pixel information in the form of descriptors that capture essential visual characteristics. By working with these compressed descriptor representations rather than raw pixel data, the system achieves accurate matching while managing computational and memory requirements
Data Source
AI summary
A method for controlling a robotic device is presented. The method includes capturing an image corresponding to a current view of the robotic device. The method also includes identifying a keyframe image comprising a first set of pixels matching a second set of pixels of the image. The method further includes performing, by the robotic device, a task corresponding to the keyframe image.