A most interesting paper from Fabrice Noreils from April 8, 2024.
Extract (verbatim)
"A humanoid robot is an extremely complex machine at the intersection of many disciplines such as mechatronics, control algorithms for underactuated dynamic systems, actuators design, energy consumption optimization, autonomous system in terms of perception localization planning and locomotion...
In addition, all algorithms must be carried out by on-board computers in a very constraint footprint.
Finally, the objective is to put these very complex, therefore fragile, machines in a demanding industrial environment to perform at least as well as a human worker within 2 or 3 years?
This is reminiscent of the enthusiasm and certainties of promoters of autonomous vehicles in the 2010s.
The question we can ask ourselves is the following: is this objective realistic or is it achievable not in 2 or 3 years but within 10 years or more? The aim of this document and its main contributions is to provide some hints by covering the following topics:
1. An analysis of 12 companies based on eight criteria described in section 3. Although these criteria are subjective, they have the benefit of being able to distinguish companies based on their maturity and approach to the market;
2. These humanoids are very complex systems facing technological challenges which are discussed in section 4; 3. Operation and maintenance are critical activities specially when humanoids are deployed at scale. This topic is covered in section 5;
4. Pilots are the last step to test the feasibility of a new system before mass deployment. This is an important step to test the maturity of a product and the strategy of the humanoid supplier to address a market. This topic is addressed in section 6."
