US EditionAI-Powered Vacuum Robots Fail Simple Tasks in New Study
Andon Labs tested six leading large language models on a vacuum robot, finding top accuracy only around 40%, highlighting significant challenges in robotic task execution.
- Last week, Beijing-based Noetix Robotics launched presales for its child-sized Bumi, selling more than 200 units and Five hundred in two days at about $1,408 each while scaling multi-factory production with nearly all in-house components.
- This year, a wave of humanoid launches in China has pushed makers toward low prices amid a robotics resurgence, as advances in AI and robotics promise more automation of tedious tasks.
- Noetix, AI research group, tested state-of-the-art LLMs on a vacuum robot and found Gemini 2.5 Pro and Claude Opus 4.1 scored 40% and 37%, while Claude Sonnet 3.5 caused a `doom spiral`, highlighting practical failures in embodied tasks.
- Noetix is scaling production toward more than 1,000 robots per month while acknowledging profit margins are quite low and completed a nearly 300 million yuan pre-B financing round.
- Physical-World data shortages have driven teleoperation and remote-control training as Objectways annotated 15,000 videos and Figure AI captured footage from 100,000 homes globally.
16 Articles
16 Articles
LLMs tried to run a robot in the real world – it didn't go well
Researchers at Andon Labs recently evaluated how well large language models can act as decision-makers in robotic systems. Their study, called Butter-Bench, tested whether modern LLMs could reliably control robots in everyday environments – particularly in carrying out multi-step tasks like "pass the butter" in an office setting.Read Entire Article
Narita International Airport Corporation and Nomura Research Institute have begun a demonstration experiment in which robots sell souvenirs unmanned at the domestic boarding gate area of Narita Airport's Terminal 3. The aim is to solve the labor shortage and improve operational efficiency. According to Nomura Research Institute, this is the first such experiment in Japan, and it will be available until December 15th.
The researchers of Andon Labs, known for their luscious experiences mixing robots and artificial intelligence, have struck again. After having entrusted the management of an ATM to a d的IA model, they have this time equipped a simple robot vacuum cleaner with several large language models (LLM) — among them ... Read more Like KultureGeek on Facebook, and follow us on Twitter Don't forget to download our free iAddict app for iPhone and iPad (link …
Coverage Details
Bias Distribution
- 75% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
