Robotic Pick Pose Estimation for Arbitrary End Effector Footprints

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing robotic picking systems struggle to accurately estimate optimal pick poses for arbitrarily sized end effectors, especially in chaotic or occluded scenes, often relying on deep learning methods that require extensive training data and prior object model knowledge, leading to suboptimal and computationally expensive solutions.

Innovation Solution

A method that models the end effector as a 2D shape with specified dimensions, estimates a pick point on an object mask using instance segmentation, determines a pick surface from neighboring points, reprojects these points to create a planar representation, and computes a yaw-oriented pick pose based on alignment with the end effector model, reducing computational cost and improving accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If deep learning methods are used to estimate pick poses, then accuracy can be improved, but computational cost and training data requirements increase significantly

Engineering Contradiction:
Improvepick pose estimation accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces complex deep learning computational systems with a simplified geometric computation system. Instead of using neural networks that require extensive training data and computational resources, the invention uses direct geometric relationships between the end effector model and object mask to calculate pick poses through coordinate transformations and alignment operations.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent extracts only the essential geometric features needed for pick pose estimation - specifically the alignment between the end effector's longer dimension and the pick surface's longer dimension. By focusing only on this critical alignment relationship rather than processing all object features through complex learning models, the system achieves accurate pose estimation with minimal computational overhead.

Inventive Principle:
Principle #2Taking out (Extraction)

2Ease of manufacture

If traditional teach-based approaches are used, then programming complexity is reduced, but flexibility and adaptability to arbitrary end effectors deteriorate

Engineering Contradiction:
Improveprogramming easeVSAvoidend effector adaptability
Core Design Contradiction:
Ease of manufactureVSAdaptability or versatility

Solution Approach 1:

The patent creates a universal pick pose estimation method that works with arbitrarily sized end effectors through geometric alignment principles. The system models any end effector as a 2D shape with specified dimensions and uses the same alignment-based computation regardless of the specific end effector geometry, making the system universally adaptable without requiring reprogramming or retraining for different configurations.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Productivity

If simple alignment methods are used, then computational speed is improved, but accuracy in chaotic or occluded scenes deteriorates

Engineering Contradiction:
Improvecomputational throughputVSAvoidpick pose accuracy in chaotic scenes
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent transforms the 3D pick pose estimation problem into a 2D alignment problem by working with the object mask and end effector model in the image plane. By computing the alignment between the longer dimensions of these 2D representations and then transforming the result back to 3D space, the system achieves both computational efficiency and accuracy even in chaotic or occluded scenes where full 3D reconstruction would be computationally expensive.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Data Source

PatentEP4592040A1System and method for pick pose estimation for robotic picking with arbitrarily sized end effectors
Publication Date: 2025.07.30 SIEMENS AG
  • EP4592040A1 patent drawingFigure 1
  • EP4592040A1 patent drawingFigure 2A~2B
  • EP4592040A1 patent drawingFigure 3

AI summary

A methodology for estimating a pick pose for an arbitrarily sized robotic end effector. The end effector is modeled as a 2D shape with specified dimensions indicative of its footprint on an object being picked. A pick point is first estimated on an object mask of a selected object, produced by performing instance segmentation on one or more input images of a scene. A pick surface is determined utilizing neighboring points around the pick point in the object mask. A set of points in the object mask, which define an extent of the pick surface, are reprojected with respect to a normal of the pick surface, to create a planar representation of the pick surface. A yaw-oriented pick pose is computed based on alignment of a longer dimension of the end effector model with a longer dimension of the planar representation of the pick surface.