Key award for robot arm technology

WAREHOUSE ROBOTICS specialist Geekplus has been recognised with a 2026 RBR50 Innovation Award for its Robot Arm Picking Station, a system that combines autonomous mobile robots (AMRs) with AI-driven robotic picking to automate one of the last major manual bottlenecks in warehouse fulfilment.

The award, presented annually by The Robot Report at the Robotics Summit & Expo, recognises the 50 most innovative companies in global robotics. It is the fifth time the Hong Kong-listed company (HKEX: 2590.HK) has received the honour, placing it alongside the likes of ABB, Amazon, Boston Dynamics and Nvidia.

The award specifically recognised the successful deployment of the Robot Arm Picking Station at a Schneider Electric warehouse in Shanghai, where the system was integrated into an existing goods-to-person (G2P) operation.

Closing the picking gap

While AMRs have made significant inroads into warehouse storage and transport operations over the past decade, the actual picking process, selecting individual items from totes or shelving and placing them into order containers, has remained stubbornly reliant on manual labour. Rapidly changing SKU catalogues, diverse product forms and the high cost of training conventional vision models for each new item have made automated picking one of the hardest problems to solve at scale.

The Geekplus Robot Arm Picking Station addresses this by pairing a robotic arm with the company’s proprietary AI foundation model, Geekplus Brain. The system uses what the company describes as zero-shot learning technology, meaning the robotic arm can accurately identify and pick items across large and frequently changing SKU catalogues without needing to be trained on each individual product. In practical terms, this removes one of the biggest barriers to robotic picking adoption: the time and cost of programming the system every time a new product line is introduced.

The RBR50 judging panel praised the way the system complements the company’s existing AMR portfolio, noting that it automates the most difficult step in warehouse workflows and moves Geekplus closer to fully autonomous, end-to-end warehouse operations.

Schneider Electric pilot results

The pilot deployment at Schneider Electric’s Shanghai facility produced notable results. According to Geekplus, the Robot Arm Picking Station achieved double the throughput of manual picking, which typically ranges between 150 and 300 pieces per hour depending on the operator. Accuracy reached 99.99%, with a perfect 100% rate recorded during testing, a significant improvement over the mispicks and omissions that are an accepted reality of manual operations.

Perhaps most significantly for warehouse operators evaluating automation options, the system was production-ready within 48 hours of deployment. Pre-trained on large-scale real-world data via the Geekplus Brain model, it required no secondary training at the site and was able to dynamically adapt to business fluctuations and packaging changes from day one.

The company’s unified All-in-One software architecture also enabled seamless integration with Schneider Electric’s existing warehouse network and production processes, an important consideration for operators looking to layer robotic picking onto brownfield sites rather than building from scratch.

What it means for logistics operators

For warehouse operators weighing up their options in robotic picking, the Schneider Electric deployment offers a useful proof point. Labour availability remains a persistent challenge across the logistics sector, and the economics of manual picking — with its variable throughput, training overheads and error rates — are increasingly difficult to justify as order volumes grow and customer expectations around accuracy tighten.

The zero-shot learning approach is particularly relevant to operations handling large or rapidly evolving product ranges, such as e-commerce fulfilment centres, third-party logistics providers and retail distribution hubs. If the technology delivers on its promise of near-zero onboarding time for new SKUs, it could significantly lower the barrier to entry for automated picking, which has historically required substantial upfront investment in vision system training.

Geekplus says it intends to expand the solution’s application across additional industries and use cases. With more than 950 companies worldwide already using its AMR solutions, the addition of integrated robotic picking represents a logical next step towards the fully unmanned warehouse operations that the industry has long discussed but rarely achieved in practice.