Autonomous Forklift Pallet Recognition for Deviation-Aware Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing unmanned forklift technologies face challenges in accurately identifying pallet types and detecting position and orientation deviations, leading to potential errors in load handling, increased costs, and reduced operation speed, especially when handling multiple pallet types without pre-attached patterns or precise distance measurements.
Innovation Solution
An unmanned forklift equipped with an image obtaining section, pallet type identification using machine learning, pallet position/shape data from a distance measuring device, and a traveling controller to perform operations based on detected deviations, allowing for accurate identification and handling of various pallet types without pre-attached patterns or extensive calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If laser range finder scanning is used to identify pallet types by detecting hole edges, then pallet type identification can be performed, but identification accuracy deteriorates when holes are obscured by stretch film, foreign substances, or adjacent objects
Solution Approach 1:
The patent introduces a camera as an intermediary detection device that captures images of the pallet opening end face. This image-based approach serves as a mediator between the laser range finder and the pallet type identification system, allowing the system to detect pallet types through image processing even when direct laser scanning is obscured by stretch film, foreign substances, or adjacent objects
Solution Approach 2:
The patent replaces the mechanical laser scanning system with an optical imaging system (camera). By substituting the laser range finder's mechanical scanning approach with a camera-based optical detection method, the system achieves more reliable pallet type identification in obscured conditions, as images can be processed to extract pallet characteristics even when physical access to hole edges is blocked
2Measurement precision
If patterns are attached to all pallets for type identification, then pallet type can be identified, but system cost increases
Solution Approach 1:
The patent enables pallets to self-identify their type through inherent visual characteristics captured by the camera. The image processing system automatically detects pallet type based on natural features of the opening end face without requiring external pattern attachments, making the system self-sufficient and eliminating additional manufacturing costs for pattern application
Solution Approach 2:
The patent replaces expensive pattern attachment systems with a low-cost camera-based detection system. The camera and image processing algorithm serve as a disposable, one-time investment that eliminates the need for recurring pattern attachment costs, making the identification system more economically viable
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 solution enables high-accuracy detection of pallet types and deviations, preventing errors and reducing operational costs by allowing the forklift to handle multiple pallet types efficiently without the need for expensive processing capabilities, ensuring safe and precise load handling operations.
Implementation Method 1
a laser range finder 18 that can be moved upward and downward together with forks 14. The laser range finder 18 performs scanning while applying laser beams to an opening end face 19a of a pallet 19
Data Source
AI summary
An image obtaining section obtains a taken image from an imaging device. A pallet type identification section has a learning model for combinations of images of a plurality of types of pallets and types of the pallets, and identifies a type of a target pallet by inputting, to the learning model, the taken image of the target pallet, which is obtained by the image obtaining section. A pallet position/shape obtaining section obtains position/shape data of the target pallet from a distance measuring device for measuring a distance to the target pallet. A pallet deviation detection section previously stores position/shape data of the pallets and performs comparison between the stored position/shape data corresponding to the identified type of the target pallet and the position/shape data of the target pallet.