Bin-picking Robot Shaking Scheme for Random Object Grasping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated systems struggle to efficiently pick up objects randomly arranged in bins due to overlapping objects and bin walls, which prevent robots from reaching prehension features, leading to incomplete or inaccurate object recognition and grasping.
Innovation Solution
A method involving a shaking scheme to rearrange objects, a routine to determine if the bin is free of recognizable objects, and a selection process for choosing the appropriate object to pick up, using a combination of vision systems and robotic manipulation to ensure the robot can grasp objects without hitting bin walls, while incorporating protection mechanisms to prevent damage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If objects are randomly arranged in a bin, then the system can handle unstructured industrial scenarios, but the robot cannot reach prehension features due to overlapping objects and bin walls
Solution Approach 1:
The system dynamically adjusts the bin orientation using tilt and shake operations. The bin can be tilted along different axes and shaken to rearrange objects, transforming the static picking problem into a dynamic one where the robot can access previously unreachable prehension features through controlled movements of the bin container
Solution Approach 2:
The system adds temporal and orientational dimensions to the picking process. Instead of attempting to pick objects from a fixed bin orientation, the system introduces multiple bin orientations (through tilting and shaking) and sequences picking operations across different time steps, allowing the robot to access objects from multiple dimensional perspectives
2Productivity
If the robot attempts to grasp objects near bin walls, then more objects can be accessed, but the gripper may hit the walls and cause damage
Solution Approach 1:
The system performs preliminary bin tilting and shaking operations to rearrange objects and create safe picking trajectories before the robot attempts to grasp. By pre-positioning objects and clearing potential collision paths through bin manipulation, the system enables the robot to reach previously inaccessible objects without risking collisions with bin walls
Solution Approach 2:
The bin itself acts as an intermediary mechanism between the robot and the objects. By tilting and shaking the bin, the system mediates the interaction, allowing objects to be repositioned and presented to the robot in safe, accessible orientations without requiring the robot's gripper to directly navigate hazardous near-wall regions
3Measurement precision
If the vision system scans all objects, then complete recognition is achieved, but overlapping objects prevent accurate detection of prehension features
Solution Approach 1:
The vision system operates dynamically in conjunction with bin tilting and shaking operations. By capturing images at multiple bin orientations and combining this visual information with knowledge of object positions from different viewpoints, the system reconstructs complete object models including previously obscured prehension features, achieving accurate detection without requiring all objects to be simultaneously visible
Data Source
AI summary
A method for picking up objects randomly arranged in a bin using a robot having a gripper for grasping the objects using prehension feature(s) on the object. The method includes a shaking scheme for rearranging the objects in the bin when no objects are recognized, when no objects are prehensible by the gripper or when the object to be picked up is not reachable by the gripper because, for example, its prehension feature is substantially facing a wall of the bin. The method also includes a criterion for determining that a bin is free of objects to be picked up and a criterion for selecting the object to be picked up first in the bin. The method also provides for a protection mechanism against damage of the objects and the robot when a recognition technique has failed in properly recognizing the object or the prehension feature on the object.