Warehousing is a complex business of operational efficiency. At DB SCHENKER, we are constantly innovating to find new ways to optimize operations and provide exceptional service. Enter computer vision – a powerful technology with the potential to change warehouse management.

The Power of Computer Vision

Computer vision goes beyond simply recording footage. Imagine a system that can automatically detect and track pallets, measure times, and provide real-time data on warehouse operations, therefore complementing the data in existing Warehouse Management Systems. This translates to an abundance of valuable data for warehousing. This technology allows DB SCHENKER to address specific operational challenges and gain valuable insights, ultimately enhancing productivity.

Computer Vision is a topic of major strategic importance because it can be scaled to all business areas and makes a significant value contribution.

Dr. Joachim Weise
SVP Global Technology and Data

The idea: To enhance transparency and provide live data, offering deeper insights about actual dock-to-stock cycle times for inbound pallet management.

Born from this was PalletVision!

PalletVision is a product of innovation born from a project initiated in 2022 at the DB SCHENKER Enterprise Lab, a collaborative venture between DB SCHENKER and the renowned Fraunhofer Institution for Material Flow and Logistics IML. This partnership brings together industry expertise and research capabilities to tackle logistics challenges and drive advancements in the field of logistics. To further improve warehousing processes, DB SCHENKER is piloting this technology on their site in Hildesheim!

This tool is powered by computer vision and additional AI methods to detect and track objects such as forklifts or pallets in DB SCHENKER warehouses. But let’s dive deeper into the AI algorithm.

How PalletVision Works

Neural network-based object detection uses a deep learning model to analyze an image and identify objects within it. The network is trained on a large dataset with labeled images, learning to recognize patterns and features associated with different objects. Objects- such as pallets, forklifts, and persons are annotated and fed to the neural network during training. During inference, the trained model processes input images and predicts the presence and location of objects within the images. To determine how long pallets are standing in the inbound area, pallets need to be tracked in a series of frames provided by the camera. Therefore, DB SCHENKER has developed a dedicated tracking algorithm for the assignment of each bounding box in the current frame to each bounding box in the next frame.

Increased efficiency: The PalletVision Advantage

PalletVision significantly enhances warehouse efficiency through automated detection and tracking of incoming pallets. Translating video streams into data, the user-friendly frontends provide critical inbound KPIs for warehouse management and put-away priorities for forklift drivers, thus ensuring process transparency and streamlined operations. Customers ultimately benefit from improved bottleneck management and Service Level Agreement compliance. DB SCHENKER leverages this solution with a robust architecture featuring bird-eye cameras and hardware installations designed for near-real-time results.

Supporting Every Level – Empowering Teams

PalletVision empowers not just the management level but also the individuals on the ground floor.

  • Shift Leaders & Management: Access live overview of inbound areas, prioritize pallets for faster processing, and generate comprehensive reports on actual dock-to-stock cycle times through the Power BI interface.
  • Forklift drivers: Gain instant transparency on priority put-away lanes, reducing the need for constant communication with shift leaders, and ultimately streamlining the workflow.

PalletVision unlocks new potentials in Warehouse Management – DB SCHENKER continues to explore the endless possibilities of computer vision across all business units to innovate operational excellence.

Published: July, 2024