Robot Grip Point Selection Using Multiple Vision Algorithms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current handling systems face challenges in accurately determining optimal grip points for robotic end effectors when picking objects from containers, especially in complex scenarios like bin-picking, due to variations in object types and environments.
Innovation Solution
A computer-implemented method that utilizes multiple grip point determination algorithms to analyze image data from detection devices, such as cameras or sensors, to select the best grip point based on various criteria, including confidence values and probability of success, allowing for precise and reliable object grasping across different objects and environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple grip point determination algorithms are used to improve selection accuracy, then the reliability of object grasping is improved, but the device complexity increases
Solution Approach 1:
The system segments the grip point determination task by employing multiple specialized algorithms, each optimized for specific object types or grasping scenarios. This allows the system to divide the complex problem into manageable parts, where each algorithm handles particular cases, thereby improving overall reliability without requiring a single overly complex algorithm to handle all situations.
Solution Approach 2:
The system changes parameters by selecting different algorithms based on object characteristics, environmental conditions, and task requirements. By dynamically adjusting which algorithm is used based on input parameters such as object geometry, material properties, and desired grasp force, the system achieves high reliability across diverse scenarios while maintaining manageable complexity through parameter-based selection.
2Measurement precision
If multiple algorithms are executed to determine optimal grip points, then the precision of grip point selection is improved, but the processing time increases
Solution Approach 1:
The system performs preliminary actions by pre-evaluating algorithms and pre-selecting appropriate algorithms based on object characteristics before the actual grasping task. This allows the system to have algorithms ready and configured in advance, reducing the time required during actual execution while maintaining high precision through careful preliminary selection and setup.
Solution Approach 2:
The system applies partial action by executing only the necessary number of algorithms based on the specific task requirements and object characteristics. Rather than always running all available algorithms, the system selectively executes only those needed for the current situation, thereby achieving sufficient precision without the full time cost of exhaustive algorithm execution in every case.
3Adaptability or versatility
If multiple algorithms are used to handle variations in object types and environments, then the adaptability of the handling system is improved, but the device complexity increases
Solution Approach 1:
The system achieves universality by designing algorithms that can handle multiple object types and environmental conditions through a unified framework. Each algorithm is designed with multi-functionality to adapt to various scenarios, allowing the system to maintain high adaptability while avoiding the need for separate specialized systems for each object type or condition, thereby managing complexity through versatile, multi-purpose algorithm design.
Data Source
Figure 1
Figure 2
AI summary
The invention relates to a computer-implemented method for controlling a handling system, the method comprising performing one or more control cycles, each control cycle comprising receiving image data representing an image of at least a section of an object to be grasped, acquired by means of a detection device, determining a target grip point on the object for the end effector, comprising analyzing the image data, generating control signals that cause the at least one robot to grasp the object at the target grip point by means of the end effector, wherein determining the target grip point comprises analyzing the image data by two or more independent grip point determination algorithms (104-1, 104-2, 104-3), wherein each of these grip point determination algorithms identifies at least one grip point candidate (106-1, 106-2, 106-3).wherein the handle candidates determined by the two or more handle-determination algorithms form a set Me of handle candidates, and selecting a handle candidate from the set Me as the target handle depending on one or more predefined handle selection criteria. The invention also relates to a handling system and a computer program product.