Robotic Object Grasping Without 6D Pose Estimation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional object grasping systems struggle with grasping objects that are overlapping, in random poses, or have deformable and varying appearances, as they rely on biased object pose detectors and traditional keypoints, leading to accumulated errors and inefficiencies.

Innovation Solution

The method determines invariant features of objects, such as keypoints and masks, to compute grasp proposals without requiring 6D object pose estimation, allowing for grasping objects in complex scenes and reducing errors by using a robotic system with an end effector and computing system to execute grasp proposals based on calculated scores.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If conventional object pose detectors and traditional keypoints are used for grasping, then the system can process simple objects, but it fails to accurately grasp overlapping, deformable, or randomly posed objects due to accumulated errors

Engineering Contradiction:

Improvegrasping accuracyVSAvoidpose estimation complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The patent extracts and eliminates the problematic 6D pose estimation step from the traditional grasping pipeline. Instead of detecting object pose and then computing grasps, the system directly computes grasp proposals from point cloud data and depth images, bypassing the source of accumulated errors. This extraction of the faulty component resolves the contradiction by improving reliability without adding complexity.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The patent inverts the traditional approach by not estimating pose first and then grasping, but rather directly computing grasp proposals from visual data. The system uses a point cloud processing pipeline that generates candidate grasp poses directly from geometric features and depth information, reversing the conventional sequence and eliminating the intermediate pose estimation step that causes errors.

Inventive Principle:

Principle #13The other way round (Inversion)

2Reliability

If the system processes complex scenes with overlapping and deformable objects, then grasping accuracy improves, but processing time and computational load increase

Engineering Contradiction:

Improvegrasping robustnessVSAvoidprocessing time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent segments the complex grasping problem into distinct processing stages: point cloud generation from depth images, candidate point selection based on geometric features, grasp proposal computation, and scoring/ranking. This segmentation allows the system to efficiently handle complex scenes by processing only relevant geometric features rather than performing full 6D pose estimation, reducing computational time while maintaining robustness.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system performs partial pose estimation by computing only the necessary grasp-related parameters (grasp pose, approach direction, contact points) rather than full 6D object pose. This partial action approach provides sufficient information for reliable grasping of complex objects without the computational overhead of complete pose estimation, reducing processing time while maintaining accuracy.

Inventive Principle:

Principle #16Partial or excessive action

3Loss of information

If traditional 6D object pose estimation is performed, then complete object information is obtained, but errors accumulate and grasp accuracy decreases

Engineering Contradiction:

Improveobject pose informationVSAvoidgrasping precision

Core Design Contradiction:

Loss of informationVSMeasurement precision

Solution Approach 1:

The patent extracts only the essential geometric information needed for grasping (surface normals, curvature, contact points, approach directions) directly from point cloud data, eliminating the need for full 6D pose estimation. This selective extraction of necessary information maintains sufficient precision for accurate grasping while avoiding the error accumulation inherent in traditional pose detection pipelines.

Inventive Principle:

Principle #2Taking out (Extraction)

4Adaptability or versatility

If the system uses invariant features and bypasses pose estimation, then grasping of complex objects is enabled, but the system requires advanced point cloud processing capabilities

Engineering Contradiction:

Improveobject type coverageVSAvoidprocessing system complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent replaces the mechanical/computational system of 6D pose estimation with a direct point cloud-based grasp proposal system. By substituting the traditional vision-to-pose-to-grasp pipeline with a vision-to-grasp direct computation approach using geometric features and depth information, the system achieves greater adaptability to various object types while managing complexity through efficient point cloud processing algorithms.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11541534B2Method and system for object grasping

Publication Date: 2023.01.03 INTRINSIC INNOVATION LLC

AI summary

A method for object grasping, including: determining features of a scene; determining candidate grasp locations; determining a set of candidate grasp proposals for the candidate grasp locations; optionally modifying a candidate grasp proposal of the set; determining grasp scores associated with the candidate grasp proposals; selecting a set of final grasp proposals based on the grasp scores; and executing a grasp proposal from the set of final grasp proposals.