Robot Grasp Control Using Graspable-Part State Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing grasping robots require time to recognize and grasp the whole three-dimensional shape of a movable target object before initiating a displacement operation, which is inefficient for quickly grasping and moving a graspable part.
Innovation Solution
A grasping robot equipped with a learned model and a lookup table that allows the robot to quickly identify and grasp a graspable part of a movable target object by using image data from an image-pickup unit, detecting its position, and controlling the grasping mechanism based on the recognized state, enabling rapid displacement operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the robot recognizes the whole three-dimensional shape of the target object, then the grasping accuracy is improved, but the time required to start the displacement operation increases
Solution Approach 1:
The patent extracts only the essential information needed for grasping (graspable part position and movable state) from the complete three-dimensional shape recognition process. By using a learned model to directly identify graspable parts and a lookup table to determine movable states from position data, the system obtains sufficient grasping information without performing full shape recognition, thus reducing processing time while maintaining grasping accuracy.
Solution Approach 2:
The patent pre-establishes a lookup table that maps graspable part positions to their movable states. This preliminary preparation allows the robot to quickly determine the movable state by simply querying the table with detected position information, rather than analyzing the complete three-dimensional shape at runtime, thereby significantly reducing the time to start displacement operations.
2Productivity
If the robot uses a learned model and lookup table to quickly identify graspable parts, then the productivity is improved, but the device complexity increases
Solution Approach 1:
The patent replaces complex mechanical shape recognition and analysis systems with a learned model (likely a neural network) that directly processes image data to identify graspable parts. This substitution of mechanical/algorithmic complexity with a trained computational model enables faster processing and higher productivity, as the model can quickly predict graspable parts and their states from images without complex geometric computations.
3Speed
If the robot grasps only a graspable part rather than the whole object, then the speed of displacement operation is improved, but the reliability of grasping may deteriorate
Solution Approach 1:
The patent applies local quality by focusing the grasping operation on the specific graspable part rather than the entire object. The learned model identifies the precise location and characteristics of the graspable part, and the lookup table provides the movable state for that specific part. This localized approach allows the robot to grasp and move objects quickly by targeting only the relevant portion, while maintaining reliability through accurate identification of the correct grasping point and its associated movement characteristics.
Data Source
AI summary
A grasping robot includes: a grasping mechanism configured to grasp a target object; an image-pickup unit configured to shoot a surrounding environment; an extraction unit configured to extract a graspable part that can be grasped by the grasping mechanism in the surrounding environment by using a learned model that uses an image acquired by the image-pickup unit as an input image; a position detection unit configured to detect a position of the graspable part; a recognition unit configured to recognize a state of the graspable part by referring to a lookup table that associates the position of the graspable part with a movable state thereof; and a grasping control unit configured to control the grasping mechanism so as to displace the graspable part in accordance with the state of the graspable part recognized by the recognition unit.