Mykolas JuraitisThis week I was experimenting with unstable robot behaviour, namely pumping a swing:
I tried various strategies like periodically turning robot on one leg, moving leg up an down aligned with swing seat or perpendicular to it. Apparently pumping a swing is quite a complicated process. In practice it is possible to achieve some results with a simple periodic motion. In the video above robot starts moving up and down with a period of 0.6s and decreases period by 2ms every swing. Motor acceleration speed also affects swing pumping so it had to be tuned.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
That's interesting that you decrease the period with each pump. I think you did at least as well as the one's using machine learning.
https://www.youtube.com/watch?v=ke6xKwXwFZU
Are you sure? yes | no
Indeed it is not obvious why decreasing period worked better than increasing in my case as period of swinging increases with larger amplitude. Maybe I was just lucky to guess some resonance causing numbers.
I did watch video you posted before starting my project to get some hints. There are surprisingly few videos of swinging robots.
I might try more scientific approach to swing higher in the future but it might require building a more sturdy swing.
Are you sure? yes | no