AI Control Automate Picking Robotics Without Training

Related Vendors

Sereact GmbH

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

With the AI-based picking robotics from Sereact, logistics service provider MS Direct is automating order picking at the port of its Auto Store for the first time. Despite over 60,000 different items, the Sereact solution now independently processes around 1,500 orders a day.

As a third-party logistics service provider, MS Direct operates highly frequented fulfillment centers for well-known e-commerce brands at several locations in Switzerland and southern Germany. The largest warehouse in Arbon (Switzerland) uses an auto-store system from Kardex with over 60,000 different items. This large variety of articles posed a major challenge for the further automation steps in the picking area: many of the numerous articles, which had different sizes, packaging and surfaces, were unknown to the first picking robot tested, which is why the desired performance and flexibility was not achieved. For this reason, picking was previously carried out mainly manually at eight of the nine ports. With the aim of establishing itself as a leading tech player in European e-commerce, MS Direct was therefore looking for a technology that could meet this challenge and process a wide variety of items without time-consuming training. The aim was to automate simple, repetitive tasks, conserve resources and at the same time be able to work productively at night.

The Picking Solution from Sereact

The collaboration with Sereact began through an existing contact: Luca Graf, CEO of MS Direct, already knew the Stuttgart-based software developer from an earlier exchange. As part of an evaluation phase, Sereact's solution was compared with other systems available on the market and ultimately selected. The main deciding factor was the potential of the AI-based control system to process new and unknown items without prior data collection. This puts Sereact technically one step ahead of other solutions.

A pick-and-place solution was used, which automates one of the nine existing auto-store ports. The software visually analyses each object in real time, recognizes relevant features such as shape, material or color and independently selects the appropriate picking strategy. The company uses an AI-controlled Vision Language Action Model (VLAM) that applies different picking strategies without manual training.

Within a few weeks, the system was installed, integrated and productive.

Marc Messerli, Vice President Fulfillment bei MS Direct

"The implementation went remarkably quickly," recalls Marc Messerli, Vice President Fulfillment at MS Direct. "Within a few weeks, the system was installed, integrated and productive—just in time for the start of the peak season." Messerli particularly emphasizes the cooperation with the Sereact team: "What convinced us was not only the technology, but also the way Sereact worked with us. Partnership-based, solution-oriented, with real enthusiasm for the topic. You could tell that this is a young team that doesn't insist on standard processes, but is prepared to break new ground together."

Productive Around the Clock

Since the commissioning of the picking robot, order picking at MS Direct has changed noticeably. The robot now independently processes around 1,500 single-pick orders a day. The solution is particularly effective in night-time processing: while the auto-store system was previously only used during the day, the robot now takes over the preparation work for the next day during the night. This significantly increases throughput and at the same time reduces the need for personnel at off-peak times.

A key objective was also achieved: simple and repetitive tasks were successfully automated. Employees are specifically deployed where human experience is required, for example for more complex order picking tasks. MS Direct also benefits from fewer errors, a reduced susceptibility to errors in the Auto Store and greater process reliability. The deployment also pays off economically: the return on investment was already achieved after around nine months—a benchmark that was further boosted by the high wage costs in the Swiss market.

"It was also important to us not to see the robot simply as a machine, but to actively integrate it into our team," explains Marc Messerli. "That's why we held a naming competition and today 'Pico' is not just a robot for many of us, but almost a colleague."

Further Expansion Planned in Arbon

With the introduction of the pick robotics solution from Sereact, MS Direct has reached an important milestone on the road to automation. The technology used has not only proven itself in day-to-day operations, but has also demonstrated its ability to reliably process a wide variety of items without prior training. This means that the company is able to remain flexible even with changing product range requirements.

"The successful collaboration with Sereact motivates us to gradually expand automated picking at other ports at the Arbon site," summarizes Messerli.

Automation is also being considered at the other locations. These are currently in various stages of technical development. As soon as the infrastructural requirements are in place there, the integration of pick robotics solutions will be examined.

MS Direct is also in discussions with Sereact regarding the possible introduction of additional technologies such as the AI-based monitoring platform Lens. This could enable even more accurate stocktaking and more efficient quality control in the future.

(kip)

Design simulation for industrial robots

How to make your production more flexible with robotics