Robbyant Unveils LingBot-Depth 2.0 and LingBot-Vision to Redefine Robotic Spatial Perception

Robbyant, an embodied AI company under Ant Group, has announced the launch of LingBot-Depth 2.0, a major upgrade to its spatial perception model, alongside the new foundational visual model LingBot-Vision. The release represents a significant advancement in enabling robots to accurately perceive and navigate complex physical environments.

Building on the original LingBot-Depth, which introduced the Masked Depth Modeling (MDM) technique to overcome depth sensing difficulties with transparent and reflective surfaces, version 2.0 dramatically expands training data and performance. Trained on 150 million samples, the new model achieves top rankings in 12 out of 16 depth completion benchmarks. In the most challenging indoor scenarios with substantial depth loss, LingBot-Depth 2.0 reduces depth error by half compared to its predecessor—lowering root mean square error from 0.132 to 0.062. The model excels particularly in situations that traditionally defeat conventional depth cameras, such as glass, mirrors, and transparent objects.

Breakthroughs in Visual Foundation Model

The performance leap is driven by LingBot-Vision, a novel visual foundation model and the first in the industry to adopt "boundary structure" as a pre-training objective. This approach delivers sub-pixel-level boundary localization and spatial structure understanding, providing the precise visual representations essential for robust spatial perception.

Remarkably, LingBot-Vision’s pre-training corpus consists of only 160 million images—an order of magnitude smaller than DINOv3. Despite this, it offers highly stable object boundary determination, enabling continuous tracking of object boundaries across video sequences. Beyond supporting LingBot-Depth 2.0, LingBot-Vision serves as a versatile foundation model capable of powering a wide range of downstream tasks.

Commercial Collaboration with Orbbec

For commercial deployment, LingBot-Depth 2.0 has received professional certification from the Depth Vision Laboratory of Orbbec, a leading provider of robotics and AI vision solutions. Real-world testing using chip-level depth data from Orbbec’s Gemini 330 series stereo 3D cameras has demonstrated marked improvements in edge clarity, object contour integrity, small object recognition, long-range depth estimation, and robustness under complex lighting and material conditions.

Within Orbbec’s newly released Robot-Free Data Collection Hardware Platform, the RGB-D EGO device will integrate a customized LingBot-Depth model optimized for data collection. Future plans include incorporating an advanced commercial version to continuously enhance missing depth completion, object edges, and spatial structures—delivering a precise and stable data foundation for embodied AI training. In collaboration with Robbyant, Orbbec will launch:

• SDK Product: Enabling edge-side deployment for robotics customers, significantly improving depth perception for systems using Gemini 330 series cameras.
• Integrated Camera: Scheduled for release by year-end, this all-in-one solution will combine "3D Camera + Spatial Perception" in a single package.

Open Source and Future Vision

Robbyant has open-sourced the model weights of LingBot-Vision via GitHub and Hugging Face. The company remains committed to building the robotic vision foundation together with industry partners, addressing critical bottlenecks in "seeing, seeing accurately, and seeing stably" in the real physical world, and accelerating the commercial rollout of embodied intelligence.

For more details, visit the project pages:
- GitHub: https://github.com/Robbyant/lingbot-vision
- Hugging Face: https://huggingface.co/collections/robbyant/lingbot-vision

About Robbyant

Robbyant is an embodied intelligence company within Ant Group, dedicated to advancing embodied intelligence through cutting-edge hardware and software. The company independently develops foundational large models for embodied AI and explores next-generation intelligent devices, aiming to create robotic companions and caregivers that enhance everyday life and deliver reliable intelligent services in key areas such as elderly care, medical assistance, and household tasks. Learn more at www.robbyant.com.

The source for this article is https://www.roboticstomorrow.com/news/2026/07/07/robbyant-unveils-lingbot-depth-20-and-lingbot-vision-to-redefine-robotic-spatial-perception/26812/.