Surprisingly Useful AI Article Enhancements
Robbyant, an embodied AI company operating within Ant Group, has released LingBot-Depth 2.0 – a next-generation spatial perception model – alongside a new foundational visual model called LingBot-Vision.
The dual release is designed to improve how robots perceive and navigate physical environments, with a particular focus on solving longstanding challenges around transparent and reflective surfaces such as glass and mirrors.
Building On Depth Sensing Foundations
LingBot-Depth 2.0 builds on the original LingBot-Depth model, which introduced a technique known as Masked Depth Modeling (MDM) to address depth sensing problems caused by transparent and reflective surfaces.
The new version represents a scale-up in both training data and performance. Trained on 150 million samples, the model has achieved top rankings in 12 out of 16 depth completion benchmarks.
In indoor scenarios involving depth loss – historically one of the more difficult use cases – LingBot-Depth 2.0 has halved the depth error compared to its predecessor, reducing RMSE from 0.132 to 0.062.
The model also shows improved performance in cases that have traditionally caused problems for conventional depth cameras, including glass, mirrors and transparent objects.
A New Visual Foundation Model
Underpinning LingBot-Depth 2.0 is LingBot-Vision, which Robbyant describes as the first model in the industry to use “boundary structure” as a pre-training objective.
The model offers sub-pixel-level boundary localisation and spatial structure understanding, providing the visual representations needed for spatial perception tasks.
Robbyant has noted that the pre-training corpus for LingBot-Vision consists of 160 million images – an order of magnitude smaller than that of DINOv3. Despite the smaller dataset, the model delivers stable object boundary determination and can continuously track object boundaries within video.
Beyond its role in supporting LingBot-Depth 2.0, LingBot-Vision is intended as a versatile foundation model capable of handling a range of downstream tasks.
Commercial Partnership With Orbbec
On the commercial front, LingBot-Depth 2.0 has been certified by the Depth Vision Laboratory at Orbbec, a provider of robotics and AI vision technology.
Real-world testing using chip-level depth data from Orbbec’s Gemini 330 series stereo 3D cameras has demonstrated improvements in edge clarity, object contour integrity, small object recognition, long-range depth estimation and robustness across complex lighting and material conditions.
Within the product matrix of Orbbec’s Robot-Free Data Collection Hardware Platform, the RGB-D EGO device will integrate a customised LingBot-Depth model optimised for data collection. Orbbec plans to further integrate an advanced commercial version to improve missing depth completion, object edges and spatial structures.
New Products In The Pipeline
In collaboration with Robbyant, Orbbec will launch new products integrating LingBot-Depth 2.0.
The first is an SDK product enabling edge-side deployment for robotics customers, designed to enhance depth perception for systems using Gemini 330 series cameras.
Related: Best Business Laptops for work & school
Related: Best Gaming Laptops
Related: Best Portable Laptop
The second is an integrated camera planned for release by the end of the year. This all-in-one solution will deliver a combined “3D Camera + Spatial Perception” package.
Open-Source Commitment
Robbyant has open-sourced the model weights of LingBot-Vision, making them available through both GitHub and Hugging Face.
The company has indicated it remains committed to building robotic vision foundations alongside industry partners, with a focus on resolving the bottlenecks of “seeing, seeing accurately and seeing stably” in real-world physical environments and accelerating the commercial deployment of embodied intelligence.
About Robbyant
Robbyant is an embodied intelligence company within Ant Group. It independently develops foundational large models for embodied AI and explores next-generation intelligent devices. The company’s stated aim is to create robotic companions and caregivers for use cases including elderly care, medical assistance and household tasks. More information is available at www.robbyant.com.
Last Updated on July 7, 2026 by Nick Ross
