Robot Gripper Selection Using Vision and Success Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Teaching robots to grip objects in industrial settings is challenging due to the need for specialized programming and expertise, and existing methods do not effectively account for industrial requirements, making it time-consuming and requiring robot experts.
Innovation Solution
A method using a robot device with multiple gripping elements, a camera, and machine learning to detect and determine the best gripping area, which involves object detection, sensor unit localization, probability determination, and storage in a database, allowing non-experts to easily teach gripping positions and select the most suitable gripper based on success probabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a graphical user interface with mouse-controlled pointer is used to program gripping positions, then the gripping position can be precisely specified, but the process becomes time-consuming and requires robotics expert knowledge
Solution Approach 1:
The system enables non-expert users to program gripping positions independently through intuitive interaction with the graphical user interface, eliminating the need for robotics expert knowledge while maintaining precise positioning capabilities
Solution Approach 2:
The graphical user interface acts as an intermediary between the user and the robotic system, translating simple mouse clicks into precise gripping position coordinates, thereby bridging the gap between user intent and precise robot execution without requiring specialized knowledge
2Loss of time
If machine learning is used to automatically determine gripping positions, then programming time is reduced and expert knowledge is not required, but industrial requirements and specifications are not taken into account
Solution Approach 1:
Industrial requirements and specifications are pre-configured into the system before the gripping task begins, ensuring that automatically determined gripping positions comply with industrial standards without requiring manual adjustment or expert intervention during execution
Solution Approach 2:
The system incorporates feedback mechanisms that verify automatically determined gripping positions against pre-defined industrial requirements, ensuring compliance while maintaining the efficiency benefits of automated machine learning-based position determination
3Adaptability or versatility
If multiple differently designed gripping elements are used to grasp different objects, then the robot system becomes more versatile, but the complexity of determining the best gripping element increases
Solution Approach 1:
The system segments the gripping element selection process by creating dedicated neural networks for each gripping element type, where each network independently evaluates its own gripping success probability, thereby simplifying the overall selection process despite having multiple different gripping elements
Solution Approach 2:
The system changes the evaluation parameter from complex multi-criteria decision analysis to simple probability comparison, where each gripping element's success probability is calculated independently and the highest probability determines the selection, reducing complexity while maintaining versatility
Data Source
Figure 1~2
AI summary
The invention relates to a method for gripping an object (10) by a robotic device comprising at least two differently designed gripping elements. The invention further relates to a computer program and a data carrier.