Pallet Pocket Detection Using Deep Learning Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated material handling systems using forklifts or pallet lift vehicles face challenges with labor availability, process efficiency, and accuracy in detecting and tracking pallet pockets, especially when load positions are variable and not accurately known beforehand, leading to slow and inefficient operations.
Innovation Solution
A system comprising multidimensional physical space sensors, a navigation system, and a command system that generates a three-dimensional representation of the pallet location space, uses deep learning for segmentation, and employs online learning to improve the detection and tracking of pallet pockets, allowing for real-time guidance of the vehicle to accurately engage with the pallet pockets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If current automated material handling systems are used, then labor availability is reduced, but detection accuracy and tracking precision of pallet pockets deteriorate
Solution Approach 1:
The system performs preliminary scanning of the pallet to identify pocket locations before the pickup operation. The sensor suite scans the pallet surface in advance, detects pocket features, and pre-calculates pickup parameters, enabling accurate detection without requiring manual input during the actual pickup process.
Solution Approach 2:
The patent replaces manual mechanical inspection and positioning with an automated sensor-based detection system. Sensors optically scan the pallet surface to identify pocket locations, replacing the need for manual measurement and positioning that would otherwise be required for accurate detection.
2Adaptability or versatility
If load positions are variable and not accurately known beforehand, then adaptability is improved, but process efficiency and accuracy deteriorate
Solution Approach 1:
The system dynamically adapts to variable load positions by performing real-time scanning and detection. Rather than relying on predetermined static positions, the sensor suite actively scans the pallet surface to identify actual pocket locations, and the control system dynamically adjusts pickup parameters based on detected positions, maintaining efficiency despite position variability.
Solution Approach 2:
The system performs self-positioning and self-detection by autonomously scanning the pallet and identifying pocket locations without external guidance. The sensor suite and control system work together to automatically determine load positions and adjust pickup parameters, eliminating the need for manual position input or external positioning systems.
3Measurement precision
If deep learning and online learning systems are implemented, then detection precision is improved, but device complexity increases
Solution Approach 1:
The system performs preliminary scanning and data collection before the pickup operation. The sensor suite gathers optical data about the pallet surface and pocket features in advance, which is then processed by the deep learning system to identify pocket locations and characteristics, improving detection precision through pre-processing.
Solution Approach 2:
The online learning system continuously improves detection accuracy by learning from each scanning and pickup operation. The system receives feedback from successful detections and uses this information to refine its detection algorithms, progressively improving precision while adapting to different pallet and load configurations.
Data Source
AI summary
A system for directing a vehicle using detected and tracked pallet pocket comprises the vehicle, a navigation system, and a command system which detect and track a pallet pocket during automated material handling using the vehicle where load positions vary and are not accurately known beforehand.