search

mechmotum / fall-probability-paper

Paper on the effect of the balance assist system on fall probability.
https://mechmotum.github.io/research/balance-assist-bicycle.html
1 stars 0 forks source link

Change to new measured parameters of the balance assist bicycle. #8

Closed moorepants closed 3 months ago

moorepants commented 3 months ago

Here is the balance assist bike without rider uncontrolled and controlled (gain 10): uncontrolled-eig-vs-speeds balance-assist-controllers-eig-vs-speeds

moorepants commented 3 months ago

Now with a rider: uncontrolled-eig-vs-speeds balance-assist-controllers-eig-vs-speeds

This shows that the balance assist stabilizes even with the rider! Which was not what we found before.

moorepants commented 3 months ago

From Marten's thesis where he identified the weave mode of the rider less bicycles we see: image compare this to the new values from our inertia measurements: balance-assist-controllers-eig-vs-speeds

Seems likely to be a better match.

moorepants commented 3 months ago

Had some error and now I think it is correct. I use the updated parameters from the linked PR on BicycleParameters as there was some error before in the fork inertia. I also wasn't applying the same controller to the with and without rider but now I am. I also changed the controller to include now look like this (as per Marten's report): gains-vs-speed

this is the new figure in the paper:

balance-assist-eig-vs-speeds

from these two inertias:

bicycle-with-geometry-mass

Two things:

  1. It does seem to stabilize the bike with rider! Which is good.
  2. I don't think it matches with the weave identification experiment. I wonder if we should lower the gain until we get a match in the frequency curve.
moorepants commented 3 months ago

Merging, will open issues to work on the rest.