Robotic Part Picking with Vision-Based Orientation Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robotic systems for picking parts from a bin struggle to determine the orientation and interlocking of picked parts, leading to errors such as accidental picking of multiple parts, shifting of parts, and incorrect placement, which can cause jams and safety concerns.
Innovation Solution
A robotic system equipped with cameras and a computing device that analyzes images to ensure only one part is picked and its orientation meets predetermined criteria, with the ability to adjust the part's position and orientation if necessary, and to return extra parts to the bin.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If contact or non-vision sensing means (force or proximity sensors) are used to detect picked parts, then the system can detect if a single part has been picked, but it cannot determine the orientation of the picked part or the nature of interlocking if multiple entangled parts have been picked
Solution Approach 1:
The patent replaces mechanical sensing means (force or proximity sensors) with a vision-based detection system using cameras and image processing algorithms. This substitution enables the system to capture visual information about the picked part's orientation and detect interlocking of multiple parts, thereby preventing information loss while maintaining detection capability.
Solution Approach 2:
The patent introduces an intermediary computational processing step that analyzes images captured by cameras. This intermediary processing extracts orientation and interlocking information from visual data, bridging the gap between simple detection and comprehensive part state assessment.
2Productivity
If the robot picks parts from the bin without verification, then the picking speed is fast, but errors such as accidental picking of multiple parts, shifting of parts, and incorrect placement occur
Solution Approach 1:
The patent implements a feedback mechanism where cameras capture images of the picked part, and image processing algorithms analyze whether the part is correctly oriented and whether multiple parts are entangled. Based on this feedback, the system can reject incorrect picks and reattempt, thereby maintaining high picking speed while ensuring reliability.
Solution Approach 2:
The patent performs preliminary verification of the picked part's orientation and entanglement state immediately after picking but before placement. This preliminary action prevents incorrect parts from being placed, ensuring reliability without significantly impacting overall productivity.
3Reliability
If the system uses vision-based detection to verify part orientation and entanglement, then picking accuracy is improved, but the system complexity increases due to additional cameras and computing requirements
Solution Approach 1:
The patent designs the vision system to perform multiple functions: detecting part orientation, identifying entanglement of multiple parts, and verifying successful picking. This multi-functionality reduces the need for separate specialized sensors for each detection task, thereby limiting the increase in system complexity while maintaining high reliability.
4Reliability
If the robot returns picked parts to the bin when orientation criteria are not met, then picking accuracy is maintained, but the picking cycle time increases
Solution Approach 1:
The patent applies partial verification by checking only the most critical aspects of part orientation and entanglement rather than performing exhaustive analysis. This partial action approach maintains picking accuracy for critical defects while minimizing the time penalty, as the image processing algorithms are optimized to detect only the most important orientation and entanglement conditions.
Data Source
AI summary
A robot system (10) for picking parts (41) from a bin (40) uses the image from one or more cameras (38) to determine if the robot gripper (24) has picked one part or more than one part and uses one or more images from one or more cameras (38) to determine the position/orientation of a picked part. If the robot (12) has picked more than one part from the bin (40) then attempt is made to return the excess picked parts to the bin (40). The position/orientation of a picked part that does not meet a predetermined criteria is changed.