Frequently Asked Questions

Why working on humanoid robots ?

Frequently Asked Questions

Questions collected following comments and questions during presentations and public science fairs.

Last updated: April 1, 2021


1. Why working on humanoid robots ?

As described in my research statement, the scientific question which drives me and many other researchers, is to discover general principles for generating motions on humanoid robots. For this we take inspiration on how human are able to perform every day behaviors such as looking for keys, playing soccer, or mounting an Ikea furniture. Once put in mathematical terms most of these problems are very difficult to solve.

2. Overview

Humanoid robot research is a very challenging robotics field of research which forces one to think differently. It is necessary to take into account all the constraints so that the robot stays balanced, do not self collide, and perform a useful action. Pushing forward the limits of those complex machines has several useful consequences for other kind of robotics.

3. Why continue to work on humanoid robots ? Boston Dynamics solved all the problems !

Until very recently humanoid robots have not yet reached the potential to perform tasks in the real world. Despite the very impressive advances from Boston Dynamics to achieve parkour, this robot is not for sale. In order to achieve this potential Boston Dynamics has developed a very powerful robot based on hydraulic based actuation. Thanks to this, the company is constantly driven the whole community to higher level. However this is posing security and energy consumption problems. Another path is to use electric based actuation. Japan with the ASIMO robot from Honda, the HRP series from AIST and Kawada, the Jaxon robots developed in the JSK Laboratory (Univ. of Tokyo), the HUBO robots from KAIST, the Toro robot from DLR have demonstrated impressive capabilities with electric based robots. Finally the newly designed Digit robot is a pretty impressive light weight biped robot. For this reason I am working on the TALOS robot.

4. Why are you trying to redo a human ?

The goal is not to redo a human. Anyone who has study how human works is amazed by its complexity and its efficiency. Few numbers are settling quickly the differences: a human has 600 muscles, usual humanoid robot have from 30 to 64 motors. The human skeleton has between 200 and 213 bones, while most of humanoid robots have 30 bones. The way the human brain is working is still the object of intensive research. My goal is to build robotic systems with some level of autonomy in their motion. Humans, and more generally Nature, are a source of inspiration. The human shape is definitely pushing roboticists outside their comfort zone.

5. I send you emails why are you not answering ?

Due to a large volume of request by emails I am not able to answer all of them. If you are looking for a PhD position or a training practice please follow the open positions that are on the laboratory website or on For practice training it is better to consider at least 5 months in our team. Please note that most of the positions I propose imply a good level in numerical optimization and software skills.

6. Open source

Most of the code I write in the frame of my work is open-source: BSD-Clause 2, LGPL, Apache or MIT. It has several advantages. It is a natural complement to the scientific papers. It provides a form of scientific reproducibility. Open source allows me to save time, and I learned a lot from other open-source projects and the talented people who wrote them. And finally the tax-payers get what they paid for :)

7. Open publications

CNRS has a policy of providing all the preprints from its employees through the French National platform HAL. You can find all the papers freely available at the LAAS HAL server