Robot End Effector Grasp Pose Refinement Using Pre-Grasp Vision

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing robotic grasping systems face challenges in determining the appropriate grasp pose for objects due to inaccuracies in end effector positioning and noisy vision data, leading to a low success rate of object grasping, especially when there are errors in traversing to the pre-grasp pose and occlusions.

Innovation Solution

The system captures end effector vision data after traversing to the pre-grasp pose and determines the final grasp pose by selecting pre-stored visual features that match the current visual features, using machine learning models to generate output for the grasp pose and predict success measures, allowing for adjustments based on similarity thresholds and kinematic feasibility checks.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of time

If the end effector traverses to a pre-grasp pose based on initial vision data, then the grasp pose can be determined in advance, but positioning errors occur due to actuator inaccuracies and calibration issues

Engineering Contradiction:

Improvegrasp determination timeVSAvoidend effector positioning accuracy

Core Design Contradiction:

Loss of timeVSMeasurement precision

Solution Approach 1:

The system captures end effector vision data after traversing to the pre-grasp pose, compares it with pre-stored visual features, and determines the final grasp pose based on the actual position. This feedback loop corrects positioning errors by using real-time vision data to adjust the grasp pose determination.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system pre-stores visual features and grasp criteria before the grasping operation. This preliminary preparation enables rapid comparison and final grasp pose determination without complex real-time computation, reducing time loss while maintaining accuracy.

Inventive Principle:

Principle #10Preliminary action

2Productivity

If the grasp pose is determined based on initial vision data, then the grasp can be planned in advance, but vision data noise and occlusions result in errors

Engineering Contradiction:

Improvegrasp planning efficiencyVSAvoidgrasp pose accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system uses end effector vision data captured after traversing to the pre-grasp pose to determine the final grasp pose. This feedback mechanism corrects errors caused by noisy or occluded initial vision data by comparing actual visual features with pre-stored features.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

Inventive Principle:

Principle #10Preliminary action

3Ease of operation

If the end effector moves along the Z-axis from the pre-grasp pose, then the grasp can be executed, but errors in the pre-grasp pose propagate to the final grasp pose

Engineering Contradiction:

Improvegrasp execution simplicityVSAvoidfinal grasp pose accuracy

Core Design Contradiction:

Ease of operationVSMeasurement precision

Solution Approach 1:

The system determines the final grasp pose based on end effector vision data captured after traversing to the pre-grasp pose, rather than simply offsetting from the pre-grasp pose. This feedback approach corrects accumulated errors by using actual visual features to determine the final position.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS11331799B1Determining final grasp pose of robot end effector after traversing to pre-grasp pose

Publication Date: 2022.05.17 X DEVELOPMENT LLC

AI summary

Grasping of an object, by an end effector of a robot, based on a final grasp pose, of the end effector, that is determined after the end effector has been traversed to a pre-grasp pose. An end effector vision component can be utilized to capture instance(s) of end effector vision data after the end effector has been traversed to the pre-grasp pose, and the final grasp pose can be determined based on the end effector vision data. For example, the final grasp pose can be determined based on selecting instance(s) of pre-stored visual features(s) that satisfy similarity condition(s) relative to current visual features of the instance(s) of end effector vision data, and determining the final grasp pose based on pre-stored grasp criteria stored in association with the selected instance(s) of pre-stored visual feature(s).