AI-Based Identification Solutions in Logistics

Scanner free logistics based on Artificial Intelligence save significant work effort and add intrinsic stability to processes

Fast AI-based barcode detection, identification and reading replaces manual scanning

Deep Learning Technology for object identification without barcodes; counting and measuring products

Perfect integration with the W2MO 3D Case Pack functionalities for optimized packing and 3D packing schemes

Flexible setup applications through all processes from inspection at inbound logistics to fast packaging at outbound

Direct quality control and documentation with images of the goods after opening / before closing the carton or container

Full integration into the W2MO Logistics Suite allows direct use of timestamps to calibrate your warehouse simulation and calculate logistics and value-added services

Highlights

Automate goods receival with a scannerless process that identifies products, gets dimensions and weight, and reads any additional information if necessary.
Speed up packing by avoiding manual scanning of products, as well as by utilizing a 3D packing scheme and having the process documented automatically.
Scan many barcodes and QR codes at one glance to accelerate the identification process.
Count many objects simultaneously using AI and computer vision technology for a safe and smooth packing process.
Measure objects during the identification process for a continuous improvement of master data availability and quality.
Train neural network for object identification using a simple process with a self-explaining interface.

Scanner free packaging

Camera and recognition box installed at work station for packing
Focus of camera on handling area — person is not filmed
Automatic identification of products and quantities by moving them through the detection zone before putting them into the box / container
Combination of barcode reading, text recognition, and object identification for different products can be done in one step
Automatic comparison with packaging list and immediate confirmation or notification of discrepancy. Visual and acoustic feedback.
Visualization of 3D packing scheme
Storing of images for later visual verification or to provide information to customers
Communication with other systems (WMS, SAP)

The whole packing process at a work station is observed by a camera with a recognition box. As only a certain “detection zone” is in focus of the camera, a person is not visible on the video stream so that there are no privacy issues. With the use of algorithms from AI and Computer Vision labels, barcodes and QR codes are decoded, text is read, and objects are identified. One of the big advantages is that several objects can be identified in one glance. An automated comparison with the packing list is the last step to this scanner free packing solution. Correct products are confirmed, for wrong items an error notification is raised. As the whole solution is fully integrated in W2MO, the 3D packing scheme can also be visualized. As the whole packing process is documented pictures can be stored for documentation and later reference.

Count and Measure

Automated counting of objects
Possibility of customer specific enhancements for special setups
Measuring of products in all three dimensions
Optional setup with integrated scale for weight measurements
Continuous improvement of master data
No additional effort for evaluating master data

Besides identifying products, the number of items for each product can also be determined. This supports the packing process but can also be applied in, for example, a production step. Special conditions, like identification of packaging material and counting of non-labelled items, can be learned by a customer-specific neural network. During these processes geodata of the products can also be measured and stored in a linked system. With an integrated scale the same applies to the weight of the product. So master data are constantly checked, extended and improved. This has a direct positive impact on all logistic processes that rely on master data like setting up pick tours, finding the right packaging material, or assigning the product to a well dimensioned bin in the warehouse.

Large scale object recognition

Possible Application: Identification Gate

Automatic detection and booking of boxes/pallets when passing a gate
Automated lighting for best conditions to read labels
No need to stop the transporter
Traceability with help of video documentation possible

Possible Application: Automated Goods Receival

Identifying and booking all visible pallets with a label
Reading labels from a large distance
Utilizing a rotary PTZ camera to find all the labels in a larger area
Zooming to every detected label and reading information

Logivations’ revolutionary identification technology can be used for automatic product recognition and direct booking. To accomplish this, identification gates can be set up at any station in the warehouse. For example, the load of a tugger train can be automatically detected and identified right when it is passing through an identification gate. It is not necessary to stop the transporter. The camera finds and recognizes all the barcodes or products / cartons and displays the linked object information either on a control screen or forwards it directly in real time via the W2MO interface to the warehouse management system for fully automatic booking of the goods. The resulting video documentation also enables complete traceability.

If a larger area needs to be observed, for instance an inbound area where the trucks arrive, PTZ (pan-tilt-zoom) cameras are used. All visible pallets and cartons of a delivery can be identified in only one step at the inbound station and will automatically be booked after the approval. Rotary PTZ cameras can be used to identify multiple objects from larger distances simultaneously. The camera recognizes labels automatically and zooms to the ideal recognition distance to read the barcode on it and moves automatically to the next label after the recognition.

Automatic goods recognition and booking bring numerous advantages. Manual efforts are reduced to the minimum, the throughput time drops considerably by performing identification while the process is ongoing, and productivity and process reliability increase enormously.

Training of Deep learning algorithms

Use of neural networks and deep learning algorithms for object identification
Recognition of products based on optical attributes without labels or barcodes
Training of algorithm with example images to learn patterns
Training of neural network can be done by user following a simple process with a self-explaining interface
Detection rates of more than 99% possible

The reading of barcodes is already included in Logivations AI-based object identification solutions and can be used directly after installation. If the objects have to be additionally distinguished or identified based on their visual characteristics without labels or barcodes, the intelligent deep machine learning algorithm comes into play.

The algorithm is trained with images of the relevant objects. To make this training process as simple as possible a rotating "learning platform" is available with which a minute is enough to take pictures of the products in the relevant perspectives. The algorithm indicates by itself whether additional images have to be taken. This revolutionary user-friendly training of new objects enables detection rates of over 99%. Even with minor damages, the products are reliably recognized after learning, since the trained images are composed of individual recognition features. This leads to secure identification, even if not all the elements are present or if they are partially hidden.