At Automate 2026 in Chicago, Orbbec unveiled a sweeping portfolio of AI-enhanced 3D vision systems designed to solve the most stubborn perception challenges on the factory floor. The headline announcement: a strategic partnership with Robbyant (Ant Group) to launch LingBot-Depth for Gemini 330 Series, an AI-powered depth filter that directly feeds vision-language-action (VLA) models with clean, reliable spatial data.
The Problem: Where Standard 3D Cameras Fail
Industrial automation is full of edge cases that defeat conventional depth cameras. Transparent objects, low-texture surfaces like plain walls, and shiny metallic materials all produce noisy or missing depth data. These failures cascade downstream—robots misgrasp, collisions occur, and production lines stall.
Orbbec and Robbyant attacked this problem at the chip level. The LingBot Enhanced Depth Filter takes precision data from Orbbec’s Gemini 330 active stereo cameras as baseline input, then applies Robbyant’s proprietary AI models to clean, denoise, and enhance the depth stream in real time.
Dual-Mode Processing with NVIDIA Jetson Integration
The filter operates in dual mode: it accepts passive stereo inputs but reaches peak performance when paired with the Gemini 330 series’ active stereo projection. Integration with NVIDIA Jetson Orin modules is seamless, and the SDK leverages CUDA and TensorRT for hardware-accelerated AI inference.
This architecture enables on-device edge depth processing—noise reduction, detail enhancement, and artifact removal happen at the camera edge, not in the cloud. The result is stable, real-time spatial awareness even in the most challenging industrial settings.
Real-World Applications Demonstrated
At the Automate booth, Orbbec demonstrated three concrete use cases where LingBot-Depth transformed unreliable depth data into dependable perception:
- Logistics seam tracking: Following carton seams on conveyor lines where repetitive brown cardboard patterns confuse standard stereo vision
- Reflective metal alignment: Assembly line tasks involving polished metal parts that scatter infrared patterns unpredictably
- Clear container grasping: Laboratory and pharmaceutical scenarios requiring manipulation of transparent vessels
Beyond LingBot: Orbbec’s Full Automate 2026 Lineup
Orbbec also showcased several other products targeting industrial-grade automation:
Configurable 3D DLP Structured Light Camera
Built for rigorous inspection and assembly verification, this camera achieves approximately 0.07 mm VDI/VDE accuracy and 0.05 mm Z-axis repeatability at 0.5 meters. Its compact, lightweight housing mounts directly onto robotic arms. The camera supports adaptive multi-profile shooting with fast, high-reflectivity, and precision modes—giving users flexibility across diverse production scenarios. It also captures sharp RGB imagery for AI-based object recognition workflows.
Gemini 435Le: Logistics Powerhouse
Orbbec demonstrated a self-driving forklift powered by the Gemini 435Le industrial stereo camera. The system produces dense, clean depth maps in dynamic environments, enabling precise forklift tine detection, ground-level obstacle identification, and collision avoidance. The 435Le’s PoE (Power over Ethernet) connectivity and IP67 rating make it suited for harsh warehouse conditions.
Gemini 305g: Ultra-Compact Wrist Camera
Designed for robot arm mounting, the Gemini 305g integrates with the NVIDIA Jetson Orin platform and features GMSL2/FAKRA connectors with minimal-loss cabling. This architecture ensures reliable, high-speed data transfer even under heavy mechanical vibration and electromagnetic interference—critical for collaborative arm applications.
Teradyne Robotics Adopts Gemini 305g for AI Training
In a significant design-in win, Teradyne Robotics integrated Orbbec’s Gemini 305g stereo 3D cameras into its UR AI Trainer system. This platform captures high-quality demonstration data for building advanced AI and VLA models, supporting physical AI developers in tackling complex manipulation tasks such as electronic component assembly.
Global Manufacturing Expansion
Orbbec reported partnerships with over 1,600 robotics companies worldwide, capturing market shares above 70% in both China’s service robot vision segment and South Korea’s mobile robot vision segment. To meet this demand, the company is expanding its manufacturing network:
- Shunde, Foshan (China): Smart manufacturing hub offering contract manufacturing and collaborative design-for-manufacturing services
- Vietnam (upcoming): RVMC facility serving as the primary hub for international shipments, mirroring Orbbec’s established Chinese production processes
With yearly output capacity surpassing 6 million robotic units, Orbbec is positioning itself as the volume supplier for the global robotics vision market.
What This Means for the Industry
The LingBot-Depth announcement signals a broader shift in the 3D vision industry. Camera manufacturers are no longer just selling hardware—they’re delivering AI-enhanced perception pipelines that integrate with large language models and VLA frameworks. By partnering directly with AI model developers like Robbyant, Orbbec is closing the gap between raw depth data and actionable robotic intelligence.
For integrators and robotics companies, this means faster deployment, fewer edge-case failures, and a clearer path from prototype to production. The era of AI-native 3D vision has arrived.
Comments are closed.