Adaptive ROI Selection for Bin Picking Near Wall Objects

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current approaches to determining regions of interest (ROI) for robotic grasping in bin picking are inefficient, particularly when objects are proximate to bin walls, leading to suboptimal grasp point estimation and performance issues.

Innovation Solution

An autonomous system uses a depth camera and processor to generate an adaptive ROI by expanding the default ROI from the bin bottom, excluding bin walls and minimizing empty regions, allowing for accurate grasp point determination on objects near or against walls.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a fixed default ROI is used for bin picking, then the system is simple to implement, but objects proximate to bin walls are excluded from grasp point determination

Engineering Contradiction:
Improvecapability to include objects near bin wallsVSAvoidROI determination complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent implements a dynamic ROI that automatically adjusts its boundaries based on detected object positions. The system transitions from a static fixed ROI to a dynamic adaptive ROI that expands to include objects near bin walls while excluding actual wall regions, resolving the contradiction between adaptability and complexity

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system uses feedback from depth image analysis to determine object positions and adjust the ROI boundaries accordingly. By analyzing the depth data and detecting object-wall relationships, the system dynamically modifies the ROI to include relevant objects while maintaining computational efficiency

Inventive Principle:
Principle #23Feedback

2Measurement precision

If the ROI is expanded to include all objects near walls, then grasp point accuracy improves, but computational load and processing time increase

Engineering Contradiction:
Improvegrasp point estimation accuracyVSAvoidROI determination time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies local quality by selectively expanding the ROI only in regions where objects are detected near bin walls, rather than uniformly expanding the entire ROI. This localized approach maintains high grasp point accuracy for relevant objects while minimizing unnecessary computational processing in empty regions

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The system performs partial expansion of the ROI only where needed to include objects near walls, rather than expanding to cover the entire bin. This partial action approach achieves sufficient grasp point accuracy without the excessive computational burden of processing the complete bin area

Inventive Principle:
Principle #16Partial or excessive action

3Area of stationary object

If bin walls are included in the ROI, then the region covers the entire bin area, but grasp computations are interfered with by wall artifacts

Engineering Contradiction:
ImproveROI coverage areaVSAvoidbin wall interference
Core Design Contradiction:
Area of stationary objectVSObject-affected harmful factors

Solution Approach 1:

The patent extracts and excludes bin wall regions from the ROI by detecting wall boundaries in the depth image and creating a masked ROI that covers the bin interior while excluding wall areas. This extraction approach maintains comprehensive object coverage while eliminating harmful wall interference from grasp computations

Inventive Principle:
Principle #2Taking out (Extraction)

Data Source

PatentEP4406709A1Adaptive region of interest (ROI) for vision guided robotic bin picking
Publication Date: 2024.07.31 SIEMENS AG
  • EP4406709A1 patent drawingFigure 1
  • EP4406709A1 patent drawingFigure 2
  • EP4406709A1 patent drawingFigure 3

AI summary

An autonomous system can include a depth camera configured to capture a depth image of a bin that contains a plurality of objects from a first direction, so as to define a captured image. Based on the bottom end of the bin and the captured image, the system can generate a cropped region that defines a plane along a second direction and a third direction that are both substantially perpendicular to the first direction. Based on the captured image, the system can make a determination as to whether at least one object of the plurality of objects lies outside the cropped region. Based on the determination, the system can select a final region of interest for determining grasp points on the plurality of objects.