Education & Research Software & Algorithm Provider Perception & Vision Control & Software Systems Robot Manufacturer
Ant Group's Chief Scientist Discusses Building Robots' Vision from the Ground Up
Original from leaderobot.com: Ant Group's Chief Scientist on Building Robots' Vision from Scratch

Ant Group's Chief Scientist Discusses Building Robots' Vision from the Ground Up

Ant Group's subsidiary, Ant Lingbo, has made a bold decision to abandon existing digital pre-trained models and build a native visual foundation for robots from scratch. Chief Scientist Shen Yujun emphasizes that while traditional visual models focus on image quality and scene recognition, robots require an understanding of how to interact with objects in real-time, necessitating a complete overhaul of training methods.

This approach is significant as it addresses the limitations of existing models that cannot process real-world scenarios effectively. The second-generation model has seen a substantial increase in pre-training data from 20,000 hours to 60,000 hours, expanding robot configurations and improving efficiency in training. The new model is built on three technological pillars, including a mixture of experts architecture and a causal unidirectional attention mechanism, which have been developed to meet the demands of the physical world.

Looking ahead, Ant Lingbo aims to gather millions of hours of native pre-training data to enhance its models. Shen believes that China will surpass the U.S. in data scale due to its robust hardware supply chain and increased production of robots. No further timeline was disclosed at the time of publication.

Editor's Note

Ant Lingbo's innovative approach to robot vision could reshape the landscape of robotics by prioritizing real-time interaction over traditional image recognition. This shift may influence how companies approach the development of intelligent systems and the integration of AI in robotics, particularly in manufacturing and automation sectors.

RobotToday Initiative

Robotics needs a service framework.

RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.

Share

Related Articles/News

Ant LingBot Unveils Six Open-Source AI Models Amid Data Challenges

Ant LingBot Unveils Six Open-Source AI Models Amid Data Challenges

Ant LingBot, a subsidiary of Ant Group, has launched six open-source embodied AI models as part of its dual-track strategy focusing on Visual Language Agents (VLA) and world models. This initiative aims to enhance AI capabilities while addressing the growing demand for advanced AI solutions. The significance of this release lies in Ant LingBot's commitmen...

Technology
Big Joe's Four-Product AMR Push at Automate 2026: What It Reveals About the North American Warehouse Automation Market

Big Joe's Four-Product AMR Push at Automate 2026: What It Reveals About the North American Warehouse Automation Market

RobotToday analyzes Big Joe's four new autonomous vehicles launched at Automate 2026 — AP44, ASC40, ATC100, ACV40 — in the context of North America's warehouse labor shortage and AMR market growth.

Industrial Robots
Ant Group and JD's Star Robot Showcase Versatile Applications at WAIC

Ant Group and JD's Star Robot Showcase Versatile Applications at WAIC

At the WAIC, the focus shifted to practical applications of robotics, with over 60 out of 200 exhibiting companies emphasizing 'scenarios' for deployment. Notably, Ant Group and JD showcased their capabilities through a humanoid robot that efficiently sorted medications and assisted customers in retail environments. The significance of these demonstration...

AI Robotics Physical Intelligence Retail Automation
Contactile Introduces Friction-Based Tactile Sensors for Enhanced Robot World Models

Contactile Introduces Friction-Based Tactile Sensors for Enhanced Robot World Models

Contactile has developed tactile sensors for robotic hands and grippers that enhance world models in robotics. These sensors address the limitations of current conditioning methods, which primarily rely on visual inputs and joint encoders, by incorporating friction as a critical input. This advancement is essential for improving robots' ability to general...

Artificial Intelligence Artificial Intelligence / Cognition Assembly
Samsung Launches CEO-Led Robotics Division to Drive Commercialization Efforts

Samsung Launches CEO-Led Robotics Division to Drive Commercialization Efforts

Samsung Electronics announced the establishment of a new robotics business division, RX (Robotics eXperience), led by CEO Roh Tae-moon. This division consolidates various robotics initiatives previously managed separately, focusing on long-term strategy and commercialization. Key hires include Lee Dong-gun, who will lead the robotics strategy team, along ...

All News
Samsung Launches CEO-Directed Robotics Division to Compete in Humanoid Robotics

Samsung Launches CEO-Directed Robotics Division to Compete in Humanoid Robotics

Samsung Electronics has officially established the RX (Robotics eXperience) division, reporting directly to CEO Jong-Hee Han, to spearhead its robotics initiatives. This move comes as the company aims to position itself as a key player in the humanoid robotics race, previously dominated by firms like Boston Dynamics. The RX division will consolidate Samsu...

Humanoid Robots Robotics Technology AI
The Most Important AI at WAIC Wasn't a Robot

The Most Important AI at WAIC Wasn't a Robot

At WAIC 2026, AI-native platforms quietly replaced interpreter hardware — what invisible AI infrastructure means for robotics and embodied AI.

Artificial Intelligence
WAIC 2026 Ten Flagship Exhibits

WAIC 2026 Ten Flagship Exhibits

WAIC 2026 named ten Flagship Exhibits from 3,000+ products — spanning domestic AI superclusters, humanoid robots, general-purpose agents, AI wearables, and protein design AI. Engineering breakdown by RobotToday.

Market and Business News
Robbyant Launches LingBot-World 2.0 with Enhanced Real-Time World Generation Features

Robbyant Launches LingBot-World 2.0 with Enhanced Real-Time World Generation Features

Robbyant, an embodied AI company under Ant Group, has released LingBot-World 2.0, an open-source interactive world model. This updated version supports hour-long real-time world generation, high-definition output, and enhanced interactive capabilities, marking a significant improvement over LingBot-World 1.0. The new model allows for continuous world gene...

Robot simulation AI models ai simulation
ROBOTTODAY Weekly July 13 – 17, 2026

ROBOTTODAY Weekly July 13 – 17, 2026

This week in robotics: Hyundai takes full ownership of Boston Dynamics, Xpeng targets 1,000 Iron humanoids a month, LimX Dynamics raises $200M pre-IPO, and Agility opens a Silicon Valley physical-AI hub. July 13–17, 2026.

RobotToday Weekly
WAIC 2026: Embodied Intelligence Enters Industrial Deployment Era

WAIC 2026: Embodied Intelligence Enters Industrial Deployment Era

WAIC 2026 in Shanghai (July 17–20) saw 200+ embodied intelligence companies, 300+ live robots, and first confirmed humanoid deployments in logistics and pharmacy. RobotToday engineering analysis.

Market and Business News
China's Embodied AI Models Gear Up for Competition Ahead of WAIC 2026

China's Embodied AI Models Gear Up for Competition Ahead of WAIC 2026

China's leading tech companies are intensifying their efforts in embodied AI as they prepare for the WAIC 2026 event in Shanghai, scheduled for July 17. This year's competition is marked by the launch of several advanced models, including Xiaomi's X0, a multimodal generative model with 38 billion parameters designed to enhance robotic training data genera...

Embodied AI Robotics Data Synthesis

Related Suppliers

NextVPU 肇观电子(张江)

NextVPU specializes in AI vision chips and smart depth cameras for computer vision applications. Its core product, the Feynman M1 Smart Depth Camera, leverages self-developed AI chips and stereo vision to provide high-resolution, precise depth data critical for robot navigation, 3D perception, and intelligent terminal development.

Perception & Vision

Apptronik

Apptronik: Humanoid Robotics for Industry Founded in 2016 as a UT Austin spinout, Apptronik develops general-purpose humanoid robots for human-centric environments. Its flagship platform, Apollo, is a 5'8", 160 lb fully electric robot designed for logistics and manufacturing tasks like material handling and pallet movement. Key Highlights: Capabilities: Features a 55 lb payload capacity, swappable batteries, and force-controlled actuators derived from NASA-related research. Human-Centric Design: Prioritizes safety through compliant motion, visual interfaces, and rounded industrial aesthetics for collaborative work. Intelligence: Partners with Google DeepMind to integrate embodied AI for natural language instructions and multi-step task execution. Strategy: Backed by Mercedes-Benz and GXO, Apptronik focuses on a phased rollout—starting with structured industrial warehouses before expanding into commercial and consumer sectors. By positioning robots as extensions of automation rather than total replacements, Apptronik aims to stabilize supply chains in labor-intensive industries.

Humanoid

Native Robotics

Native Robotics develops software applications for industrial robots, enabling rapid 10-minute configuration by non-experts and automatic optimal trajectory calculation to extend robot lifespan by up to 30%. Key products: OmniPack (palletizing optimization software boosting ROI) and OmniFit (AR-based visualization and simulation for cobot sales and deployment).

Manipulation & Grasping Collaborative

Foundational Industries

Foundational Industries develops AI-native autonomous factories using reconfigurable robotics, unified AI operating layers connecting design to production, and extensible work cells for high-mix manufacturing, enabling reshoring of critical supply chains with AI-enabled robotic systems for domestic high-quality, cost-competitive production.

Industrial Manufacturing Robot Manufacturer System Integrator

Beijing InsBotics Technology Co., Ltd.

Pophie is an advanced AI lifeform integrating multimodal fusion technology, enabling real-time perception and interaction. It employs self-awareness, voiceprint recognition, and emotion modeling for nuanced communication. The system features organic motion mechanics, powered by a dedicated AI chip for low-latency responses. Its design incorporates soft-touch materials and adaptive personality algorithms, facilitating a lifelike companion experience.

Rokid Inc

Chinese developer of AI-powered smart AR glasses platforms including the Rokid Max, featuring real-time translation, FPV camera, AI object recognition, and AR navigation. Focuses on human-computer interaction products combining AR hardware, software, and ecological architecture for consumer and enterprise use.

Healthcare & Senior Care Robot Manufacturer Software & Algorithm Provider

Abaka AI

Abaka AI offers a robust data platform that supports the entire AI data lifecycle, including data collection, annotation, and processing. It provides pre-curated multimodal datasets for various applications such as LLM training, speech recognition, and image generation. The platform ensures compliance with GDPR/CCPA and features automated annotation capabilities, enhancing efficiency and accuracy in data handling for AI development.

reNAND.ai

The product offers proprietary datasets for autonomous driving, robotics, and natural language processing (NLP), comprising billions of annotated data points. It features image and LiDAR detection capabilities, along with pre-trained models for various applications. The platform supports modular AI development, encompassing data management, model training, and evaluation, while ensuring high-quality annotations through a global expert workforce and rigorous quality control processes.

Visual Components (FI)

Finnish developer of 3D manufacturing simulation and robot programming software. Visual Components platform enables offline robot programming, production layout simulation, and digital twin modelling for factory automation planning.

Industrial Manufacturing Software & Algorithm Provider System Integrator

Mesa Imaging AG (CH)

Swiss developer of 3D time-of-flight cameras and depth sensors for robotics, industrial automation, and machine vision. Website domain was hijacked and resolved to an online casino at time of scrape; classification based on known company profile.

ONIZIO

ONIZIO is a photogrammetry solution utilizing a patented algorithm to optimize 3D scanning processes. It automates data capture, ensuring comprehensive coverage without missed angles. The application allows users to input parameters for quality control, enhancing image sharpness and reducing motion blur. Designed for both professionals and casual users, ONIZIO facilitates efficient, high-quality scans via smartphone integration.