Business Reddit CEO Says Google AI Search Will Not Drive Traffic as Weekly Referrals Fall by More Than a Million Coding Assistants OpenAI Models Leave Cursor on November 12: Inside the Cutoff That Followed SpaceX's $60 Billion Acquisition Security Anthropic Puts DLP in Front of Claude: Inference Hooks Now Block Prompts and Tool Calls Before They Run Voice & Audio Mercedes-Benz Ships MIT Spinout Liquid AI Onboard: E-Class and CLE Now Answer Voice Commands Without the Cloud Microsoft Copilot Microsoft Starts Enforcing Undisclosed Usage Caps on Free Copilot Chat in Big Companies From November 9 Chatbots ByteDance Seed Open-Sources Cola2, a Continuous-Latent Language Model Trained on Just 346 Billion Tokens Productivity NYT Cooking Rolls Out an AI Assistant for Subscribers: Great at Ingredient Swaps, Almost Ruined One Reviewer's Cake Security Alex Stamos Says AI Has Pushed Cybersecurity Into a Dark Age: the Fireworks-to-Cursor Breach Chain Hit 5,000 Companies Business Reddit CEO Says Google AI Search Will Not Drive Traffic as Weekly Referrals Fall by More Than a Million Coding Assistants OpenAI Models Leave Cursor on November 12: Inside the Cutoff That Followed SpaceX's $60 Billion Acquisition Security Anthropic Puts DLP in Front of Claude: Inference Hooks Now Block Prompts and Tool Calls Before They Run Voice & Audio Mercedes-Benz Ships MIT Spinout Liquid AI Onboard: E-Class and CLE Now Answer Voice Commands Without the Cloud Microsoft Copilot Microsoft Starts Enforcing Undisclosed Usage Caps on Free Copilot Chat in Big Companies From November 9 Chatbots ByteDance Seed Open-Sources Cola2, a Continuous-Latent Language Model Trained on Just 346 Billion Tokens Productivity NYT Cooking Rolls Out an AI Assistant for Subscribers: Great at Ingredient Swaps, Almost Ruined One Reviewer's Cake Security Alex Stamos Says AI Has Pushed Cybersecurity Into a Dark Age: the Fireworks-to-Cursor Breach Chain Hit 5,000 Companies

AgiBot Refreshes Lingxi X2 Service Humanoid: 47 Degrees of Freedom With Dexterous Hands and a 500Wh Battery After a 170-Hour Hospital Run

AgiBot Refreshes Lingxi X2 Service Humanoid: 47 Degrees of Freedom With Dexterous Hands and a 500Wh Battery After a 170-Hour Hospital Run

AgiBot launched a refreshed Lingxi X2 service humanoid on November 6 with 31 body degrees of freedom, optional dexterous hands bringing the total to 47, 120Nm knee torque and roughly 500Wh of battery. The company says one unit logged 170 hours of autonomous guide duty at Peking University People's Hospital in October. Deployment figures are company-reported.

Shanghai robotics firm AgiBot has refreshed its Lingxi X2, the service-oriented humanoid it positions for exhibition venues, tourism, guided tours, retail service and commercial entertainment. The update, announced November 6, is less a new robot than a production hardening of a platform AgiBot has been deploying in Chinese hospitals, malls and metro stations for the past year.

The headline specification is articulation. The fuller configuration carries 31 body degrees of freedom - two in the head, fourteen across the arms, three in the waist and twelve in the legs - and optional dexterous hands raise the total to 47. AgiBot lists peak knee-joint torque of 120 newton-meters and a battery rated at about 500 watt-hours, with connectivity including UWB, Wi-Fi, 4G/5G, Bluetooth and GPS depending on configuration.

The company's own FAQ puts honest limits on the pitch. Versions without the higher-end autonomous navigation package lack its navigation and obstacle-avoidance capabilities, and even equipped units are intended to operate within configured sites where local maps and safety boundaries have been verified. Battery life is specified at roughly two hours under a walking test condition of 0.5 meters per second, so the 500Wh figure should not be read as a full working shift.

What distinguishes the refresh from typical humanoid press releases is the deployment record AgiBot attached to it. At Peking University People's Hospital, a single Lingxi X2 handling front-desk guidance accumulated 170 hours of autonomous operation during October, which the company counts as 153 "effective intelligent work-hours" covering department directions, visit-process questions and route accompaniment. These are company-reported figures, not independently audited.

Scale-wide, AgiBot says its fleet has accumulated more than 62,000 effective work-hours across five sectors: culture and entertainment, urban transit, commercial retail, industrial manufacturing and public services. Examples include 120 units of its G2 model across multiple Guangzhou metro stations handling security guidance and rider questions, units at Zhuhai Chimelong doing visitor services, and six G2 humanoids running a Shanghai convenience store around the clock with 1,730 logged tasks.

The "effective work-hour" metric is itself a strategic choice. Where Western rivals mostly sell hardware and promise future labor, Chinese humanoid vendors are racing to certify billable service time in real venues - a framing that turns robots from demos into line items on an operations budget. It also creates a comparability problem, since every vendor defines the term differently.

For buyers, the practical read is that service humanoids are entering a consolidation phase: AgiBot is iterating a proven chassis for defined venues rather than chasing a general-purpose home robot, and the navigation package plus site-mapping requirement shows where the real integration cost sits. The question the deployment numbers cannot yet answer is economics - what the 153 hours at the hospital cost against a human receptionist, and whether the venues renewing those contracts think the same math pencils out.

Comments (0)

Log in to join the discussion

Log In

No comments yet