By ODDO BHF AM
In this new monthly Fund Insight, we want to deal with a particular variation of artificial intelligence, rich in myths but above all in societal progress: the emergence of the humanoid robot.
On 30 September 2022, on the occasion of their AI Day, Tesla teams presented their prototype humanoid robot, Optimus. With a human silhouette, this robot is capable of walking on a flat surface, moving like a human being, carrying and moving heavy loads (such as a large parcel); allowing Tesla to assert the following value proposition: to be able to mass-produce (millions of units) and sell for less than 20,000 USD each, a humanoid robot capable of performing, initially, simple but tedious tasks for the human community.
The functional progress represented by this technological breakthrough by Tesla must be put into perspective. Until now, the mass production of robots has been limited to non-human shaped robots that were confined to the closed circuits of: a) factory automation (Japanese companies such as Fanuc, Yaskawa or American companies such as Teradyne , for example); b) mass consumption (hoovers from the company iRobot, which is in the process of being bought by Amazon , for example). They were therefore not intended to operate in open environments designed for humans. These new humanoid robots therefore open up an unprecedented field of application: that of interaction with human beings (even if in Tesla's presentation we understand that this second floor of the rocket is still at the prospective stage).
While the first attempts to build and sell humanoid robots (Honda's Asimo robot, Softbank's Pepper robot) have not been successful, caution is still required. In this case, one can nevertheless intuit in the case of Tesla, and in view of its mass success in the electric vehicle (and that which we see emerging in the autonomous vehicle), that the chances of success on this humanoid robot initiative are much higher than those of previous attempts (including the controversial initiative, but the most advanced to date, of the company Boston Dynamics).
A technological feat made possible by the pooling of numerous areas of expertise from the automotive sector
The design and manufacture of the Optimus robot is based on several families of expertise in which Tesla has already demonstrated its global leadership in its core business of electric and autonomous cars :
Artificial intelligence: One of the key success factors of a deep neural network is the amount of data (images, videos) that will allow it to learn (or "train"). In this respect, Tesla and its fleet of several hundred thousand vehicles on the road equipped with autonomous driving systems already has the best algorithm of its kind in the world. The American group was able to capitalise on this advantage by reusing for its humanoid robot the same self-piloting neural network as that of the autonomous vehicle, but which it was able to re-train for the specific needs of its robots (object identification, obstacle detection).
Actuators and electrical components which are key components of the electric vehicle. In the application framework of this Optimus robot, Tesla's strengths will be the optimization of the robot's energy consumption by three key factors: a) the centralisation of the battery pack in the thorax of the humanoid robot with the integration of all the electrical, sensory and electronic functions on a single printed circuit; b) the assumed constraint of the sensory and mobile capacities of the robot's management (only what is necessary); c) the optimisation of the robot's time (big difference with the human being). All the components needed to manufacture its robots already have dedicated production lines in the manufacturer's Fremont (California) factory, with maximum reuse of technologies already used in the group's core business.
The reuse of CAD and Simulation software solutions used in the automotive business for the robot division. Thus, crash test simulations were replaced by simulations of the robot's fall, but also of the optimisation of the actuators (there are 28 on the robot in total and one of them is anecdotally capable of lifting a load of half a ton, such as a grand piano, for example) or of all the robot's motor situations.
The societal benefits of such robots: “AI for Good”
Tesla's vision is to make a (mass-produced) robot that is useful for as many tasks as possible in everyday life. If the first demonstrations (the Optimus version 1) concern handling within the framework of the company, we can easily imagine how these new agents will be able to secure and improve our society in the fields of, for example, assistance to the elderly persons, the handicapped persons or even the protection and security of children. The prospects in terms of entertainment are also important (the human silhouette and motor skills of these robots offer a lot of potential here).
Although these first robots are not yet equipped with conversational functions, we have to imagine that in the near future they will be. In this respect, the functional spectrum that conversation can provide is very broad: combating the isolation of the elderly persons, vulnerable people or people with disabilities.
The next step will be to equip these robots with brain functions and emotions, opening the way to various applications, particularly in the artistic field and content creation.
Ultimately, these robots will probably one day be endowed with a conscious artificial intelligence and these robots will then become even more humanoid than robots.
Finally, from a more general economic and social perspective, the arrival of these new economic agents could shift the contours and definition of GDP, creating a more abundant and secure society in terms of making products and services available to the greatest number.
