What is it about?
The Nao robot is an humanoid robot, which has to balance itself when walking on two legs. It is hard to get the dynamics of walking right. The Unreal Engine has an underlying PhysX engine, which allows to model the dynamics in great detail. This also means that many parameters of the PhysX to tune the dynamics right. In this study we tuned those parameters and compared the resulting walks to the movement of the real Nao robot, which gives a high fidelty simulation.
Featured Image
Photo by Pat Krupa on Unsplash
Why is it important?
Simulation is essential in robotics, to perform experiments under reproducable circumstances and enable to allow the training of machine learning algorithms with enough variation. Yet, this is only possible when the simulation is realistic enough.
Perspectives
I like this research not only because the end-result is impressive (showing how realistic standing on one leg is simulated). It is also a great example of well designed study, with starts with simulating the falling of a simple weight to get the gravity right, and adds in a few steps complexity (joints, limbs) to get at the end a working model consisting of 21 joints connected in 5 kinematic chains (head, arms, legs).
Dr. Arnoud Visser
Universiteit van Amsterdam
Read the Original
This page is a summary of: Validation of the dynamics of an humanoid robot in USARSim, January 2012, ACM (Association for Computing Machinery),
DOI: 10.1145/2393091.2393128.
You can read the full text:
Contributors
The following have contributed to this page
