In warehouses operating with a WMS system and visual monitoring systems, partial information about business events and CCTV footage is passively collected. In my opinion, this creates a kind of “data island graveyard” that does not support anti-fraud activities. Only when an incident occurs does the “investigation” begin, which most often relies on reviewing system logs and surveillance footage. It is important to emphasize that the greatest threat comes from gangs of warehouse employees connected with security staff, transport personnel, and other company departments.

Such organized groups can operate within companies for years, exposing them to multimillion-dollar losses. Interestingly, these companies usually have already invested in CCTV camera systems, WMS software, radio networks, mobile terminals, and less often in RFID-based radio identification.

‍

We decided to unlock the power inherent in these individual elements. Our anomaly detection module uses standard, readily available technologies that require no additional investments, such as:

• WMS system logs containing all tasks with basic information about location, user, goods, order times, and operations, systematically stored;
• Historical WMS data that allow identifying indices most prone to theft, employees with the shortest tenure, freight companies, and contractors reporting the most complaints;
• Warehouse terminals equipped with the Android system collecting movement data (using accelerometers);
• Beacon beacons installed in the warehouse enabling precise micro-location of devices and users;
• IP-based CCTV cameras, whose footage is analyzed in real time. Current image recognition and ML algorithms allow highly accurate identification of objects such as humans, carts, and pallets.

‍

‍

Based on properly processed data collected from independent sources, the anomaly detection module can identify abnormal events such as: the appearance of pallets and carts in warehouse areas where no tasks were generated; violations of virtual zones within the warehouse; concentration of personnel movement unrelated to warehouse operations. Such events are automatically detected by the system, generating alerts with an assessment of potential threat levels. Thanks to this data management organization, personnel involved in warehouse security can practically react in real time to various threats. The system also has “self-learning” capabilities.

‍

In warehouses where pallets or even entire shipments are the units of theft, the system can be additionally armed with other independent data sources, such as:

• Induction loops at loading docks detecting carts;
• RFID gates — radio identification technology for goods not only speeds up warehouse processes but can also be used cost-free to detect fraud.

‍

IThere are additional mechanisms that can be tailored to the individual needs of a company and industry profile. I believe most of us have encountered many attempts to steal from and cheat companies. I think such pathological and parasitic behaviors should be condemned and fought against by all available means. In our WMS system, we also strive to manage this area..

‍

‍

‍

Additional information about the latest technologies for enhancing security in logistics can be found in the article:

A COMPUTER VISION-BASED APPROACH FOR STORAGE LOCATIONS OCCUPANCY DETECTION USING DEEP LEARNING