According to Yann Lecun, the head of AI at Meta, ‘the decade of robotics’ is currently opening up.
Yann Lecun is not the only one to predict this; one need only look at the last two keynotes of the head of Nvidia. He presents his latest graphics cards for AI with humanoid robots on stage. This is no coincidence.
Fundraising in this field is also an interesting marker. One example is Figure AI, a manufacturer of humanoid robots that has just raised $675 million.
But what does Yann Lecun say? He says that AI will benefit robotics, but also that robotics will also benefit AI. It’s a positive feedback loop that will create a new paradigm in the world of AI! Indeed, he emphasises the limitations of current generative AI, which lacks an understanding of the physical world, the famous common sense that we expect from someone or something intelligent. Another company has also recognised this and has made it its name: Physical Intelligence, a Californian company (yet another) that has just raised 400 million dollars. Their objective? Unlike traditional robotics (industrial robotics), which is often focused on specific and repetitive tasks in controlled environments, Physical Intelligence is working to create a more adaptive AI for real-world interactions. This is the definition of service robotics, a form of robotics that adapts to its environment, unlike industrial robotics, where it is the environment (the production line) that adapts to the robot.
The challenge for this company, and for all companies working in service robotics (the current standard bearer being humanoid robotics, as Elon Musk has made it a totem with his robot Optimus), is for robots to understand and adapt to their environment. These will be robots with a broader spectrum of action and therefore more useful. In the circle of robotics engineers, a well-known maxim is ‘it is (relatively) easy to make a robot, it is much harder to make it do something useful’.
We are at a tipping point. Computing power and algorithms enable us to envisage such adaptive robots.
That is what we do at Running Brains Robotics. We build surveillance robots, but above all, we programme our own algorithms for autonomous navigation and our perception algorithms so that our robots can fulfil increasingly complex and therefore increasingly useful missions. We help to address the issue of staff shortages, the protection of on-site personnel and the reduction of risks on industrial sites. Our robots also provide greater knowledge of these sites.
We therefore totally agree with Yann Lecun and that is why our teams are as involved in robotics as they are in artificial intelligence. We train our perception algorithms and use reinforcement learning so that our robots learn and always evolve towards greater utility value for our customers.
CEO of Running Brains Robotics
