Enhancing Warehouse Automation with ToF Camera Technology for AGVs

With the continuous improvement of warehouse automation, the design and application of unmanned forklifts also face higher standards. Unmanned forklifts (AGVs) must possess higher levels of stability, safety, and intelligence to ensure they can accurately identify the position of goods and efficiently complete picking and placing tasks while minimizing the risk of collisions.

ToF CAMERA Technology for Addressing the Challenges of Manual Handling and Automated AGV Collaboration

Background of the East China Lithium Battery Warehouse Project

A lithium battery automated warehouse in South China has deployed multiple AGVs to optimize the handling process. However, the daily operations of the warehouse inevitably involve manual handling of goods, leading to delays in updating the system with the current status of storage locations. This lag can result in AGVs receiving incorrect operation instructions, reducing operational efficiency and potentially causing safety risks.

To address this issue, the project plans to introduce an advanced storage location status recognition system to comprehensively monitor the status of storage locations and provide real-time feedback to the WMS system. This ensures that AGVs perform picking and placing tasks more efficiently, stably, and safely.

1. Single-point LiDAR: Single-point LiDAR emits a single laser beam to the surface of an object to form a point. This distance measurement method can easily overlook gaps between boxes or pallets, leading to the identification of the current storage location as empty or lacking goods, resulting in stacking accidents.
2. RGB Camera: When detecting target areas through deep learning, the presence of objects outside the training set in the storage location can cause false detections and incorrect storage location status information. Additionally, for storage locations that require stacking, the RGB camera lacks height information, making it difficult to arrange stacking tasks.
3. Ultra-wide-angle Fisheye Camera: Edge distortion is significant, posing challenges for model training and prediction accuracy. It also requires additional server costs with GPU support.
4. ToF CAMERA: Provides three-dimensional data and color information of storage locations, with built-in computing power, eliminating the need for external industrial computers.

Detailed Introduction of ToF CAMERA Technology

The ToF CAMERA's 3D vision storage location status recognition solution, based on RGB-D cameras, monitors the occupancy status and stacking height of each storage location in real-time, providing detailed data support for mobile robots to ensure precise picking and placing operations. This solution integrates AI deep learning technology, placing the recognition algorithm on the camera end for real-time monitoring, minimizing the likelihood of misjudgments.

• Comprehensive Monitoring: Accurately identifies storage location status through the collection of three-dimensional data and color information combined with AI technology.
• Easy Deployment: The storage location status recognition algorithm is embedded in the camera, reducing deployment and maintenance costs.
• Flexible Communication: Supports various communication methods such as TCP/IP, UDP, and HTTP, and reports data to the control system in real-time using JSON format.
• Increased Efficiency: Provides real-time storage location information, assisting the scheduling system in quickly and accurately allocating tasks.

Ultimately, the entire warehouse completed the deployment of hundreds of sensors, providing comprehensive real-time monitoring of storage location status.