Robot Grasp Pose Estimation with Circular Anchors and Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot systems for object manipulation face challenges in accurately calculating grasping positions and postures due to the complexity of learning rotation angles and the high calculation burden associated with generating and classifying numerous rotated candidate boxes, which affects the efficiency and accuracy of grasping operations.
Innovation Solution
The system employs a feature calculation unit that uses a neural network with an encoder-decoder model to fuse intermediate features, a region calculation unit that utilizes circular anchors to reduce calculation complexity, and a GC calculation unit that converts parameters to express the handling tool's position and posture, enabling more efficient grasping by focusing on high-scoring regions and reducing parameter count.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If numerous rotated candidate boxes are generated and classified to calculate grasping positions and postures, then the accuracy of grasping calculation is improved, but the calculation burden and device complexity increase significantly
Solution Approach 1:
The patent extracts only the essential features needed for grasping calculation from the full image data. Instead of generating and classifying numerous rotated candidate boxes, the system selectively processes only relevant regions and features, removing unnecessary computational steps while maintaining grasping accuracy.
Solution Approach 2:
The patent segments the image processing task into distinct stages: feature extraction, candidate region generation, and grasping calculation. By dividing the process and focusing computational resources on critical segments, the system reduces overall complexity while preserving measurement precision.
2Measurement precision
If numerous rotated candidate boxes are generated and classified to calculate grasping positions and postures, then the accuracy of grasping calculation is improved, but the processing time and productivity decrease
Solution Approach 1:
The patent performs preliminary feature extraction and candidate region identification before the actual grasping calculation. By preparing and filtering candidate regions in advance, the system reduces the computational workload during real-time operation, improving processing speed while maintaining accuracy.
Solution Approach 2:
Instead of processing all possible rotated candidate boxes, the patent applies partial action by focusing only on high-probability candidate regions. This selective processing approach maintains grasping accuracy while significantly reducing the number of calculations required, thereby improving operational efficiency.
3Adaptability or versatility
If rotation angle learning is performed to determine grasping posture, then the adaptability to different object orientations is improved, but the learning complexity and computational overhead increase
Solution Approach 1:
The patent applies local quality by focusing learning resources on specific local features and regions that are most relevant for determining grasping posture. Instead of learning all possible rotation angles globally, the system learns locally adapted features for different object orientations, reducing model complexity while maintaining adaptability.
Data Source
AI summary
According to one embodiment, an object manipulation apparatus includes one or more hardware processors functioning as a feature calculation unit, a region calculation unit, and a grasp configuration (GC) calculation unit. The feature calculation unit serves to calculate a feature map indicating a feature of a captured image of grasping target objects. The region calculation unit serves to calculate, on the basis of the feature map, a position and a posture of a handling tool by a first parameter on a circular anchor in the image. The handling tool is capable of grasping the grasping target object. The GC calculation unit serves to calculate a GC of the handling tool by converting the position and the posture indicated by the first parameter into a second parameter indicating a position and a posture of the handling tool on the image.