Robot Vision Control Using 3D Point Clouds and Complementary Color
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot systems struggle to accurately distinguish objects from their backgrounds in camera images, especially when the object and surface have similar colors or when the surface is dirty or has a mesh pattern, leading to misrecognition and potential operational failures.
Innovation Solution
A controller that acquires three-dimensional point cloud data and corresponding camera images, sets a non-object region with a complementary color to the object, and uses this information to accurately recognize the object's position, enabling precise operation control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing is used to distinguish objects from backgrounds, then the system structure remains simple, but object recognition accuracy deteriorates when objects and surfaces have similar colors or when surfaces are dirty or have mesh patterns
Solution Approach 1:
The patent transitions from two-dimensional image processing to three-dimensional measurement by introducing a depth sensor. The depth sensor captures distance information to create a depth map, enabling the system to distinguish objects from backgrounds based on height differences rather than color or texture alone. This dimensional change allows accurate object detection even when objects and surfaces have similar visual characteristics.
Solution Approach 2:
The patent segments the imaging space into object regions and background regions by comparing depth information. The depth map is processed to identify regions where height exceeds a threshold value, separating objects from the background surface. This segmentation approach enables precise object localization without being affected by color similarity or surface patterns.
2Ease of manufacture
If color-based object detection is used, then the detection process remains simple, but misrecognition occurs when objects and surfaces have similar colors
Solution Approach 1:
The patent adds a depth dimension to object detection by using a depth sensor to capture height information. Instead of relying solely on color-based two-dimensional image analysis, the system uses three-dimensional depth data to identify objects. This allows the detection process to remain relatively simple while dramatically improving reliability by detecting objects based on their height above the surface rather than color differences.
3Measurement precision
If image filtering is applied to enhance object visibility, then object distinction improves, but the original image information is altered affecting subsequent processing
Solution Approach 1:
The patent segments the depth map into object and background regions based on height thresholding. By comparing each pixel's depth value against a predetermined threshold, the system creates a binary segmentation that clearly distinguishes objects from the background. This segmentation approach enhances object visibility without altering the original depth information, preserving all measurement data for subsequent processing.
Data Source
AI summary
A controller for controlling an operation of a machine or a robot for handling an object by includes: a sensor signal acquisition unit that acquires a signal from a sensor that outputs three-dimensional point cloud data and a corresponding camera image of an imaging subject; a region acquisition unit that acquires a region having a height lower than a predetermined value; a non-object region setting unit that sets, as a non-object region, a region of the camera image corresponding to the acquired region; a complementary color setting unit that sets a color of the non-object region to a color complementary to a color of the object; an object position recognition unit that recognizes a position of the object relative to the imaging subject from the camera image; and an operation control unit that controls the operation of the machine or the robot using at least the recognized position.