Transformable Bin-Picking Box for Object Detection Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing bin-picking systems face challenges in accurately detecting and identifying target objects within randomly stacked and overlapped objects, especially when objects have reflective surface materials, leading to errors in estimating location and pose.
Innovation Solution
A bin-picking system with a transformable bin-picking box and a supporting unit that adjusts its alignment, combined with an image acquisition unit using a visual sensor and lights to extract silhouette and edge information, facilitating the detection and rearrangement of objects to improve accuracy in locating and posing the target object.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If objects are randomly stacked in a bin-picking box, then the system can handle diverse object arrangements, but detection accuracy of target object location and pose deteriorates due to overlapping and light reflection
Solution Approach 1:
The bin-picking box is made transformable rather than fixed. The supporting unit can move vertically to transform the box from a compressed state to an expanded state, dynamically adapting the box configuration based on detection results to resolve overlapping objects and improve visibility for accurate detection
Solution Approach 2:
The system performs preliminary detection of object locations and poses before picking. When detection accuracy is insufficient due to overlapping, the system first transforms the bin-picking box to separate objects, then performs detection again, ensuring accurate location and pose estimation before the actual picking action
2Device complexity
If a fixed bin-picking box is used, then the system structure is simple, but detection accuracy deteriorates when objects are overlapped and light reflection occurs
Solution Approach 1:
The bin-picking box transitions from a static fixed structure to a dynamic transformable structure. The supporting unit enables vertical movement to expand or compress the box, allowing the system to maintain structural simplicity while achieving improved detection accuracy through configuration transformation
3Measurement precision
If the bin-picking box is transformed by moving the supporting unit, then detection accuracy improves by rearranging objects, but the operation time increases due to additional movement steps
Solution Approach 1:
The system uses detection feedback to determine whether bin-picking box transformation is necessary. The control unit analyzes detection results and only triggers supporting unit movement when detection accuracy is insufficient, avoiding unnecessary transformation operations and minimizing time loss
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system enhances the accuracy of detecting and identifying bin-picking target objects by minimizing light reflection impacts and rearranging objects for clearer imaging, thereby improving the bin-picking process in manufacturing and assembly tasks.
Implementation Method 1
if the surface material of objects such as plastics has a characteristic of reflecting light, it is very difficult to identify and detect location and pose of a target object
Implementation Method 2
an image acquisition unit configured to take an image of the bin-picking candidate objects
Data Source
AI summary
Disclosed are a bin-picking system and a bin-picking method. A bin-picking system includes a transformable bin-picking box; a supporting unit configured to support the bottom part of the bin-picking box and be movable upward and downward; and a control unit configured to change an alignment of at least one bin-picking candidate object by transforming the bin-picking box by controlling a movement of the supporting unit if no bin-picking target object is detected from the at least one bin-picking candidate object existing inside bin-picking box and placed on the supporting unit to facilitate detection of a bin-picking target object and bin-picking.


