Robot Tire Picking Using 3D Point Clouds in Unknown Arrangements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tire picking systems struggle with efficiently picking tires from unknown arrangements in heterogeneous bulk loads, as they require specific hardware setups and knowledge of tire dimensions, which limits their adaptability to unpredictable environments and reduces efficiency in handling diverse tire sizes and orientations.
Innovation Solution
A system combining vision techniques and unsupervised learning to reconstruct 3D point clouds of tire arrangements, allowing a robot with a pivotable gripper to identify and pick tires based on minimal surface information, enabling stable gripping of tires regardless of their size or orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional vision systems with CAD model matching are used to pick tires, then the system can achieve precise tire identification and positioning, but the system requires prior knowledge of tire dimensions and a controlled environment with specific hardware setups, reducing adaptability to unknown arrangements
Solution Approach 1:
The patent replaces traditional mechanical vision systems that rely on CAD model matching with a neural network-based system. The neural network processes point cloud data from laser scanners to directly identify and locate tires without requiring pre-programmed tire models or controlled environmental conditions, enabling operation in unstructured, real-world tire storage environments.
Solution Approach 2:
The system changes the fundamental parameters of the vision system by transitioning from deterministic CAD-based matching to probabilistic neural network-based recognition. This allows the system to handle variable tire arrangements, orientations, and occlusions that cannot be captured by fixed geometric models, thereby improving adaptability to unknown configurations.
2Reliability
If specialized hardware setups with fixed facilities are used for tire picking, then the system can achieve reliable tire detection and handling, but the system complexity increases and the system cannot adapt to significant variations in the setup
Solution Approach 1:
The neural network-based system serves multiple functions: it detects tire presence, determines tire orientation, estimates tire dimensions, and identifies gripper contact points all within a single integrated framework. This universal approach replaces multiple specialized sensors and processing systems, reducing hardware complexity while maintaining or improving reliability across diverse tire configurations.
3Measurement precision
If model matching methods are used to detect tires in point cloud data, then the system can identify tire positions accurately, but the method requires all objects to be identical within one scale factor and largely visible, limiting its applicability to heterogeneous bulk loads
Solution Approach 1:
The system dynamically adapts to each tire's unique characteristics by using the neural network to process individual point cloud representations. Rather than requiring all tires to match a fixed model, the system learns and adjusts to variations in tire size, shape, orientation, and occlusion levels, enabling accurate detection across heterogeneous bulk loads with diverse tire configurations.
Data Source
AI summary
A tire picking system (100) performs a process for picking one or more tires stored in an unknown arrangement and for which a target location must be realized. A gripper (108) and a robot (102) including a gripping device (104) supported by a pivotable elongated arm (106), the gripping device extending from the elongated arm to a free end (104a) where the gripper is disposed, form part of the system (100). An image processing module applies data representative of the physical environment around the robot (102) to a deployed neural network in order to determine one or more parameters of a target tire (P*); the robot is set in motion based on the determined parameters of the target tire, so that the gripper can pick a target tire (P*) selected by the system (100) from among the stored tires.