Artificial Intelligence AI controls industrial trucks in real-time

Related Vendors

Linde Material Handling GmbH

FAULHABER Drive Systems

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

Wuxi InfiMotion Technology Co., Ltd.

In the optimization of material flow processes, warehouse managers will receive increasing support from artificial intelligence (AI) in the future. Linde Material Handling (Linde-MH) demonstrated how this product vision works and is gradually becoming a reality in a tech showcase at the Logimat trade fair in Stuttgart (Germany).

The solution is being developed based on Nvidia's powerful AI platform "Omniverse," which will use a digital twin to capture, process, and analyze vast amounts of process data in warehouses in real-time. Only in January, the parent company, the Kion Group, announced a large-scale collaboration with AI leader Nvidia and the digitalization experts from Accenture to elevate industrial automation to a new level.

In an extension area of the Linde-MH exhibition stand, a scenario was presented typical for the "warehouses of tomorrow": manual and automated industrial trucks working in perfect harmony with the help of innovative AI technology. Operators of large fleets particularly benefit from this, as intelligent hardware and software combined with immense computing power ensure that every process in the warehouse becomes transparent and orders can be completed faster, safer, and more flexibly through continuous simulation. "Machine learning and neural networks will make the warehouse more efficient. Throughput increases, manual and automated fleets can be optimized, and personnel used more efficiently. Overall, this leads to significant cost savings for companies," forecasts Ulrike Just, member of the management board of Linde-MH. "As one of the technology and innovation leaders in our industry, we have begun developing solutions based on AI. We are breaking new ground to improve the competitiveness of our customers and make their material flow future-proof. We are planning pilot projects with major clients, as the associated investments will particularly pay off for them."

Real-time localization creates transparency

Linde-MH's strategy initially focuses on networking manual industrial trucks. The intralogistics company is currently developing a real-time localization system that works both inside warehouses and outside buildings. It allows seamless tracking of each vehicle's location, leveraging innovative "low infrastructure" ultra-wideband technology. Drivers receive navigation instructions through the warehouse or updates on new or amended driving orders via an intelligent display. The system combines location with vehicle data, such as steering angle. This way, the route can be adjusted in real time if there is excessive traffic on driving routes, leading to potential time loss.

At some point, it only works with AI

However, the coordinated collaboration and route optimization between manual and automated industrial trucks—in the second step—place significantly higher demands on the computing power of computers due to the increasing complexity. "If 100 or more vehicles are to be coordinated, this is only possible with a superior intelligence and hardware capable of processing such large amounts of data," explains Ron Winkler, Managing Director of the Digital Business Unit at Linde-MH. "This is where the AI of the Nvidia Omniverse platform comes into play. A digital twin of the warehouse is created, a virtual 1:1 image of physical reality." In this digital twin, simulations can be calculated in fractions of a second. Either to optimize driving routes and the optimal coordination of AMR and manual industrial trucks, or to achieve optimizations in existing warehouse layouts.

The special feature: Solutions for changing demands or problems in the warehouse, such as current order or inventory changes, traffic density in certain warehouse areas, obstacles or overhanging loads, are captured in real-time, simulated in the digital twin, and recalculations are fed back to the vehicle control—for example, ensuring that a delayed arriving truck is unloaded by the nearest forklift equipped accordingly. To achieve this, all physical data of industrial trucks (such as engine performance, steering angle) and infrastructure (such as rack spaces, driving paths, machine times) are digitally stored in the Nvidia Omniverse platform. In this virtual space, information permanently comes from sensors, intelligent vehicle and infrastructure cameras, warehouse management software, and vehicle controls is processed.

Simulation is the key

The tracking of load carriers, AMR, manually guided vehicles, as well as the monitoring of the load and storage zones, is handled by intelligent camera systems on the infrastructure and on the manual and automated vehicles. Their images are directly interpreted and processed by the AI.

The showcase at the trade fair stand demonstrated this specifically: A forklift driver brought goods to the staging area with his Linde electric forklift. From the defined handover area, a fully automated Linde reach truck took the pallet to transport it into the warehouse. To seamlessly document and track materials and goods on the Omniverse platform, the manual forklift's mobile intelligent vehicle camera automatically takes a picture of the load when picking up the pallet and stores it in the system. At the same time, it detects people and obstacles and immediately adjusts the vehicle's behavior accordingly. The stationary cameras in the warehouse provide information about the occupancy of storage locations to the system and also register potential collisions with people and can initiate a reduction in vehicle speed.

But what happens if the forklift driver does not place the pallet exactly on the designated spot, as an AGV normally requires? Through the stationary cameras, the digital twin knows that the pallet has been set down and has passed the pick-up order to the Linde L-Matic core. Thanks to the intelligent camera of the fully automated AGV, the AI recognizes the tilted pallet and searches for a solution—in this case, the appropriate approach for picking up the load. Similarly, if boxes shift or a part of the load overhangs, these issues do not escape the stationary cameras. In this case, the AI would conclude that the goods should not be picked up by an AGV. The Linde L-Matic core stays put and can be assigned another transport order. Instead, the AI calculates which manually operated vehicle is nearby to take over the task.

"With the configuration of a digital warehouse image, any conceivable infrastructure and fleet configuration can be simulated in 3D optics and tested for its efficiency," explains Ron Winkler. "The AI can be continuously trained and refined. This creates the conditions for a warehouse universe that proactively solves challenges and continuously improves."

* Heike Oder is Head of Trade Press at Linde Material Handling GmbH in 63743 Aschaffenburg (Germany), Tel. +49 6021 991277, heike.oder@linde-mh.de

Smart-Factory

Access solutions for the Smart Factory - Industry 4.0