Log-Ident investigates markerless visual identification methods for logistics objects in warehouse environments. Departing from conventional identification architectures that rely on fixed camera infrastructure and extrinsic markers such as data matrix codes, the project pursues a decentralized and infrastructure-minimal approach in which pallets and, where applicable, their loads are identified directly at the point of handling based on intrinsic visual object features, e.g. using the wood grain as a visual feature.

At the center of the project is the development of a modular camera-based perception add-on for industrial trucks and mobile robotic platforms, including AGVs. The envisaged system combines object detection, image-based vehicle guidance, and visual identification in an integrated architecture that enables reliable object recognition without the need for additional warehouse-side sensing infrastructure. This permits identification functionality to be embedded directly into operational handling processes, including pickup, transport, and transfer operations.

From a research perspective, the project places particular emphasis on the comparative evaluation of camera configurations and machine learning approaches for industrial deployment scenarios. To this end, Log-Ident investigates alternative hardware and algorithmic configurations, generates annotated datasets for realistic warehouse environments, and assesses the robustness and transferability of markerless identification methods under practical operating conditions. The integration of interfaces to ERP and WMS environments further enables the synchronization of visual identification results, e.g. for object validation, position updates, and the digital association of load carriers and transported goods.

The methodological and technical approach will be validated through a physical demonstrator implemented on a comparable mobile platform equipped with the necessary sensing, power supply, and communication components. Beyond technical validation, the project will provide datasets, software modules, documentation, and generic interfaces as open-source resources. In this way, Log-Ident aims to support knowledge transfer into industrial practice and to lower implementation barriers for small and medium-sized enterprises seeking to adopt AI-based computer vision technologies in warehouse logistics.
 

This project is funded by the Federal Ministry for Economic Affairs and Energy.