The pallet detection system works with time-of-flight technology (ToF) and aims to increase productivity in logistics processes and to improve the effectiveness of automated guided vehicles (AGVs).
Forklift trucks and pallet trucks are part of everyday life in intralogistics when loading and unloading trucks with pallets. The optimisation of warehouses and many storage and retrieval processes are also hardly conceivable without these powerful vehicles. “As a rule such vehicles are still manually operated”, tells Sebastian Vögele from the customer solutions team at ifm. With a system for automated pallet detection based on our ToF cameras, we want to make such processes more cost-effective, reduce the error rate compared to manual operation and improve the overall equipment effectiveness of companies.” Some AGVs are already equipped with systems that are designed to detect the pockets in which the forks of the transport vehicles must be positioned. Normally, laser lines are used in these cases, but they require a relatively long time for a reliable position detection. Robotics developers at ifm have therefore taken a closer look at this task and, based on the in-house ToF cameras, have created a fully-fledged system that is superior to conventional approaches. It is called pallet detection system (PDS).
Evaluation time below 1 s

The 3D camera, on the basis of PMD technology, detects scenes and objects at a glance in three dimensions.
The PDF system is based on the ToF cameras of the O3D series. With this technology, it is possible to measure the distance of each individual object point to the sensor and thus capture 3D images of objects. The developers use this feature for their pallet detection system. They installed a ToF camera slightly above and between the forks of an autonomous truck that while in operation moves between 1 and 2 m to the next pallet to be picked. With triggering, the ToF camera captures a complete 3D point cloud of the pallet, which is then filtered via the PDS algorithm to eliminate noisy or unnecessary pixels. As a result, the PDS system outputs the degree-of-freedom pose of the pallet, showing the exact pallet coordinates in x, y and z directions as well as any rotation around the vertical axis or horizontal tilting. The PDS communicates this data by wire via an Ethernet interface or CAN to the vehicle’s control system, which then ensures that the load is picked up correctly. The x, y and z positions of the two pockets are also calculated in detail to ensure that the pallet is correctly picked up by the transport system. “This data is provided with a confidence factor that represents the deviation to an ideal pallet”, explains Vögele. The user can define limits for a reliable identification and thus determine the conditions for a manual intervention depending on the user’s individual requirements.” With evaluation times of usually less than one second, the PDS system is considerably faster than comparable systems and meets the usual requirements of warehouse processes in most applications. However, this task is not trivial. In warehouses there are often situations that are difficult to predict, e.g. due to pallets that have been in use for a long time and therefore no longer correspond 100 % to the ideal geometric condition, or due to stored goods that are slightly shifted or rotated in their location. Even films that are often used to wrap pallets and the goods stored on them for protection can become loose, making it difficult to detect the correct position. The PDS system is currently designed for use on pallets with two pockets, which are widely used in the industry.
Successfully used

Autonomous forklifts equipped with the PDS system drive up to a distance of between 1 and 2 m from the pallet to be gripped before a ToF camera takes a 3D image of the pallet.
Vögele sees manufacturers of AGVs who are to integrate the system into their vehicles as target customers for the PDS system: “Our competence covers both the hardware in the form of the ToF camera and the software part used to evaluate the measured results. The results achieved so far show that the PDS system is very good at the detection of pallets and calculation of their exact position. However, we do not integrate the systems into the vehicles, this job is the responsibility of the AGV manufacturer or a system integrator.” The system has already proven its suitability in practical use several times, says Vögele: “The PDS system works very reliably at many customers worldwide. At a large customer in the USA, for example, over 300 self-driving robots are in operation using the ifm system with excellent results. The system also checks the planned placement positions for pallets and ensures with its high repeatability and low downtime that the goals set are reached: The overall cost-effectiveness of the company has increased and, in addition, the number of errors has been reduced considerably compared to manual operation”.
Picture credits: