Geek+ Redefines Warehouse Automation with RoboShuttle V5 and Embodied AI Breakthroughs

In an exclusive conversation with Robotics Business News, to Carlos Sun, Head of Embodied AI Products and Solutions, and Yixin Bai, Senior Product Manager at Geek+, unveil the vision behind RoboShuttle V5—a next-generation autonomous picking solution designed to eliminate inefficiencies, address labor shortages, and enable scalable, high-speed fulfillment. Powered by the Geek+ Brain and advanced multi-vision intelligence, the system marks a major leap toward fully unmanned, data-driven warehouse ecosystems.

RoboShuttle V5 is positioned as a new industry standard—what key industry challenges were you aiming to solve with this launch?

The first is manual & semi-Automated inefficiencies. Low picking throughput and technical bottlenecks that limit productivity. Second, there is labor & continuity: chronic staffing shortages impacting operational consistency. Third, we see autonomous scalability as a major problem that many companies deal with. Their ability to scale fulfillment through unmanned systems often remains limited. And finally, deployment velocity is a huge challenge we aim to address. Balancing system complexity with the need for rapid deployment is a pain point we see across the industry that our solutions aim to address.

How do you see fully autonomous picking systems reshaping warehouse operations and supply chains globally in the coming years?

Fully autonomous picking systems are set to revolutionize warehouse operations and global supply chains by dramatically increasing productivity and speed, reducing labor dependency, and enabling smarter, data-driven operations.

Early adopters will gain significant competitive advantages in their operations. Overall, autonomous picking promises to make supply chains faster, more efficient, and highly adaptable, positioning companies to respond effectively to market fluctuations and operate sustainably on a global scale.

What are the core innovations behind RoboShuttle V5 that enable true end-to-end autonomous picking without human intervention?

The core breakthrough of the RoboShuttle V5 solution lies in its Geek+ Brain-powered unmanned picking station, which effectively bridges the "last mile" of automated warehousing:

• "Plug-and-Play" Intelligence: Leveraging deep pre-training on massive SKU datasets, the unmanned picking station possesses exceptional universal perception. This enables deployment and go-live within 48 hours of arrival, eliminating tedious parameter tuning and significantly shortening the time-to-market.

• Self-Evolving Picking Capabilities: The station achieves fully automated, high-precision picking for outbound operations. Most importantly, it is self-adaptive: as customers introduce new SKUs, the Geek+ Brain utilizes advanced generalization algorithms for real-time recognition. This removes the need for secondary modeling or manual intervention, ensuring uninterrupted production.

• From "Goods-to-Person" to "Goods-to-Robot": By combining the high-density storage of the RoboShuttle V5 with the precision of unmanned picking, the system creates a full-link, closed-loop solution. This not only liberates human labor from repetitive tasks but also enables 24/7 "lights-out" operations, building a truly resilient supply chain.

Can you elaborate on how zero-shot learning and the multi-camera vision system improve accuracy and adaptability across diverse SKUs?

By capturing visual data from multiple angles simultaneously, the multi-camera system constructs high-precision 3D images, significantly enhancing spatial intelligence. Much like human vision, the Geek+ Multi-Vision System utilizes passive sensing to acquire real-time item data within bins, generating detailed 3D point cloud models within the Geek+ Brain system.

This provides the following capabilities:

• Advanced Spatial Understanding: The system goes beyond geometric mapping to provide robust spatial reasoning. Even in cases of severe occlusion or chaotic stacking, it can accurately infer the complete spatial pose of an object.

• Multi-Dimensional Semantic Recognition: Simultaneously, Geek+ Brain identifies key semantic attributes, including material, texture, and surface characteristics.

• Intelligent Strategic Output: This multi-dimensional data feeds into strategy models for real-time simulation. It directs the robotic arm to select the optimal grasp path and force, ensuring high picking accuracy and environmental adaptability even when handling vast quantities of unknown SKUs.

• Rather than memorizing specific product names, the Geek+ Brain embodied intelligence engine enables zero-shot learning by training the algorithm to recognize universal physical properties — delivering high generalization across diverse product attributes.

RoboShuttle V5 delivers up to 2x efficiency gains and 99.99% accuracy—what does this mean in practical terms for warehouse operators?

We achieved shorter order fulfillment cycles and enhanced peak-season scalability. This effectively doubles the warehouse's throughput capacity within the same timeframe.

We reduce "hidden costs" by streamlining the order verification process, significantly lowering return-related expenses, and improving inventory precision.

How does the platform’s plug-and-play and modular architecture help businesses deploy automation faster and scale more efficiently?

The platform features a modular design for simple physical assembly on-site, eliminating the need for extensive secondary development.

The system has pre-integrated software. By leveraging universal models pre-trained on massive real-world datasets via Geek+ Brain, the system is essentially model-free. With no need for customer-side training or labeling, it offers plug-and-play functionality and 48-hour rapid deployment. New SKUs require no additional training, drastically simplifying the go-live process.

It also features flexible scalability. This enables seamless expansion without production halts. You can add workstations or robots on the fly to increase capacity with zero downtime.

Which industries or use cases are you prioritizing for RoboShuttle V5 adoption, and why?

The RoboShuttle V5 was designed to address complex logistics environments where "labor-intensive operations" meet "high-precision, high-efficiency requirements."

It is widely applicable across E-commerce, Retail, Apparel, 3PL, Healthcare, and Manufacturing. We are currently prioritizing deployment in the following industries and use cases:

• Pharmacy: Given the extreme requirements for accuracy, traceability, and strict regulatory compliance, the RS V5 is an ideal fit. Its ability to handle boxed, bottled, and small-sized items ensures a higher pickable rate. With advanced recognition and grasping capabilities, it achieves ultra-high precision, mitigating the risk of medication errors.

• E-commerce & Retail: These sectors face high SKU turnover and volatile order volumes. Powered by Geek+ Brain’s generalization algorithms, the V5 enables zero-shot learning, allowing it to identify new items instantly without long-term modeling. Its high-efficiency picking allows facilities to easily manage massive SKU counts and seasonal peaks.

• 3PL (Third-Party Logistics): Managing multi-tenant inventory requires high flexibility and a move away from rigid fixed assets. Beyond its picking prowess, the V5’s decoupled architecture and modular design allow for the flexible addition or removal of robots and workstations to adapt to shifting business needs.

Apparel & Manufacturing: The robotic picking station is adept at handling soft packaging and irregularly shaped items. Its 24/7 unmanned picking capability makes the V5 solution a perfect match for specific workflows in apparel and general manufacturing.

Looking ahead, how will Geekplus continue to evolve its robotics and AI capabilities to support the next phase of intelligent logistics?

Geek+ remains committed to defining the next generation of intelligent logistics through the deep evolution of embodied intelligence.

• From Point to Line: Closing the Loop Following the release of our robot arm picking station in October 2025, we have conquered the "final centimeter" of outbound automation, achieving a fully automated, closed-loop workflow from storage to sorting.
• From Specialized to Universal: The leap to Gino 1 In February 2026, we launched Gino 1, our first embodied intelligence robot built for warehouses. Also powered by the Geek+ Brain, Gino 1 integrates a decade of deep industry domain know-how with superior mobility and operational flexibility, making it adaptable to diverse scenarios—from transport and picking to complex multi-task coordination.
• Future Vision Driven by the continuous evolution of the Geek+ Brain, we are moving beyond individual device automation toward a "Full-Scenario Unmanned Warehouse" ecosystem. The synergy between Gino 1 and our robotic workstations marks our transition from functional robotics to Universal Embodied AI, empowering the next decade of smart logistics.