Paris-based Bleu Robotics has raised €5 million in pre-seed funding to speed up the deployment of humanoid robots in factories. Contrarian Ventures and OVNI Capital co-led the round, with participation from Kvanted, WEPA and investor Charlie Songhurst, alongside angel investor Lukas M. Ziegler.
The company is building AI software that teaches humanoid robots new factory tasks from as few as 10 demonstrations. Instead of training a robot on large datasets for each job, the system identifies the goal and constraints behind an operator’s demonstration and reproduces the task. Operators teach the robot directly at the workstation, without writing code, and Bleu is working toward a full deployment process that takes less than one hour.
The software is designed to work across different humanoid platforms, so robots can be retrained as production needs change or when they move to another location. Bleu says its first deployments are already underway with automotive manufacturers, which it did not name, and with WEPA, a European hygiene-paper producer.
The company also claims more than 99% accuracy on production tasks while using up to 100 times less data than alternative approaches. Those figures come from the funding announcement and were not accompanied by published testing methodology, so they should be read as company claims rather than verified results. Bleu was founded by Benoît Berkoukchi, who previously managed assembly-line teams at Renault, and Jean-Baptiste Mouret and Serena Ivaldi, research directors at France’s INRIA institute.
Humanoid deployment is one of the hardest problems in industrial automation, and open research shared via Hugging Face is helping the whole field move faster. As multimodal AI such as Gemini improves at understanding physical scenes, factory robots stand to become far more adaptable.
Source: The AI Insider
