The precise measurement of freight pieces is a central challenge in logistics. Especially with irregularly shaped packaging, as they frequently occur in the manufacture of machines and systems, conventional manual methods quickly reach their limits. Inaccurate measurements lead to inefficient space utilization, higher transport costs and ultimately also to a poorer environmental footprint. The solution: With advanced computer vision technology, often also called AI Vision, the measurement of freight can be standardized and carried out precisely.
Computer Vision revolutionizes logistics
To understand the importance of computer vision (CV) for logistics, it's worth looking at how this technology works. CV is a subfield of artificial intelligence that enables computers to capture, analyze, and interpret visual information from the real world. In doing so, the technology simulates the functioning of human vision by identifying objects, recognizing patterns and analyzing them.
In the example of freight measurement this becomes particularly evident: While a human, when manually measuring complex three-dimensional objects, often has difficulty accurately capturing all protrusions and overhangs, a CV system captures the object using multiple cameras from different perspectives simultaneously and creates an exact three-dimensional model.
3D measurement – an innovative solution to a complex problem
A project by Telekom MMS with the beverage machinery builder Krones
exemplifies how the measurement of freight pieces can be mastered and computer vision can be usefully applied. Krones faced the task of precisely measuring its often multi-meter-long, packaged and irregularly shaped machine parts from plants for transport. Achieving the necessary accuracy with the previous manual measurement was especially time-consuming for overhead work. To simplify and speed up this process, Telekom MMS developed an AI-assisted system for automated 3D measurement, based on three core components:
- Three LiDAR cameras (Time-of-Flight): They emit light pulses and measure the time it takes for the reflected light to return to the sensor. This yields highly precise 3D images of the freight pieces.
- Edge computing with NVIDIA Jetson: The processing of sensor data takes place directly on-site on a platform designed specifically for AI applications. This enables fast response times and minimizes data transfer.
- AI algorithms: Specialized software processes the sensor data and creates exact three-dimensional models of the freight pieces, including all relevant dimensions.
Learning from others: Challenges and approaches to solutions at Krones
The implementation of a new (AI) system naturally brings some challenges. One of the biggest at Krones was reliably capturing particularly long packaging over 5.50 meters. This required careful calibration of camera positions and the development of
specialized image-processing algorithms.
Also the integration into existing logistics processes was demanding. The system had to fit seamlessly into the workflows and communicate with existing IT systems. In close collaboration, the IT experts together with the logistics specialists at Krones developed a solution that integrates optimally into the workflow.
Another challenge was to ensure measurement accuracy even under difficult conditions. In logistics halls there is often busy operation, which can cause camera vibrations. Here, the development of compensation algorithms helped to offset such disturbances, but also the rigid mechanical mounting and the protection of the sensitive sensors.
This is how the implementation succeeds
For companies looking to deploy a similar solution, the following success factors have emerged:
- Thorough analysis of the as-is situation: Which processes does the company want to optimize? Where are the biggest savings potentials? What technical prerequisites exist?
- Step-by-step implementation: Start with a pilot project in a manageable area, collect experiences and continuous optimization, gradually expanding to additional areas
- Employee involvement: early briefing and training of employees, involvement of users in development, building internal know-how
- Technical integration: coordination with existing IT systems, definition of interfaces and data formats, consideration of security aspects
Those factors lay the groundwork for a successful introduction and sustainable
use of the CV solution.
The future of logistics is AI-optimized
The rapidly advancing development in the field of AI-assisted logistics promises even more automation and integration in the future - from full integration into Warehouse Management Systems, through automatic damage detection and predictive analytics for transport planning, to the expansion with augmented reality for visualization. The solution implemented at Krones exemplifies how AI-powered systems can transform logistics processes.
The key to success lies in the combination of precise sensor technology, powerful algorithms and well-thought-out integration into existing processes. For companies, there is an opportunity to sustainably optimize their logistics processes and gain a competitive advantage. It is important to consider both the technological aspects and the organizational and human factors. Intelligent, connected systems will shape the future of logistics. Companies that address this development early and implement corresponding solutions will benefit in the long term.
The author
Martin Wunderwald is an AI strategist at Telekom MMS, guiding companies in the DACH region in the value-oriented introduction of AI, GenAI and computer vision and bringing results into production; in addition with focuses on agent AI and AI infrastructure (cloud and on-prem) for sovereign, compliant solutions. Telekom MMS supports companies with around 2,150 employees at ten locations
