Anaxi Labs Crowdsources 20-Step Human Video Data for Robot Training
Updated
Updated · Computerworld · May 25
Anaxi Labs Crowdsources 20-Step Human Video Data for Robot Training
1 articles · Updated · Computerworld · May 25
Anaxi Labs is building a crowdsourced pipeline of egocentric videos showing people performing tasks, aiming to supply robotics companies with human-scale training data that co-founder Kate Shen says better matches real-world robot work.
YouTube clips and simulations are no longer enough, Shen said, because physical AI needs far more repeated interaction data for each scenario than internet video can provide.
This summer, the startup plans to launch a data collection and annotation app, while also working with industrial regions covering construction, logistics and factory-floor tasks.
The dataset spans about 20 common industrial actions and growing household use cases, with annotations expanding from segmentation and contact points to step-by-step reasoning and failure-recovery cases.
Shen argues the approach addresses a core infrastructure gap in physical AI and, for now, is being pulled mainly by factories facing labor shortages or dangerous jobs.
Can Anaxi's real-world data outcompete NVIDIA's synthetic data factories in the race to train truly intelligent robots?
Anaxi is crowdsourcing robot training data. How will it navigate the ethics of user privacy and fair data compensation?
Will teaching robots 'why' they perform tasks finally close the gap between mechanical mimicry and autonomous intelligence?
Anaxi Labs and CMU Redefine Robot Training: 50% Task Success Gains Through Crowdsourced Human-Scale Data and a Programmable Data Marketplace
Overview
Anaxi Labs is leading the way in crowdsourced robot training by launching a dedicated global application and forming a strategic partnership with Carnegie Mellon University. These developments expand Anaxi Labs’ reach and reinforce its commitment to cutting-edge research in physical AI. The partnership with CMU not only strengthens research and development but also highlights Anaxi Labs’ unique approach to collecting and using real-world data for training robots. By combining innovative technology with strong academic collaboration, Anaxi Labs is setting new standards for practical, real-world applications of artificial intelligence.