Skip to content

Conversation

@efernandez
Copy link
Member

This PR is on top of #23

This adds:

  1. Changes the linear covariance model into a quadratic one (on the wheel position increment). With a quadratic model, the twist covariance doesn't change with the control frequency, because it cancel the dt^2 division of the Jacobian. I want to allow the user to select between the previous and the new model (WIP).
  2. Adds a wheel_resolution param to the new covariance model, which avoids zero covariances when the robot doesn't move and makes the model more realistic. Note that this param is a constant, so the covariance would vary with the control frequency. If == 0.0 it has no effect, and the covariance (with the quadratic model) won't change with the control frequency.

@ayrton04 @afakihcpr @servos

@efernandez efernandez force-pushed the compute_twist_on_publish branch 2 times, most recently from 228b683 to 3f844af Compare January 29, 2016 05:35
@efernandez efernandez force-pushed the change_covariance_model branch 3 times, most recently from 8a44c4d to e264c15 Compare January 29, 2016 05:48
@ayrton04
Copy link

👍 LGTM

@efernandez efernandez force-pushed the compute_twist_on_publish branch 4 times, most recently from 6d190f0 to 088839f Compare February 1, 2016 18:58
@efernandez efernandez force-pushed the change_covariance_model branch from e264c15 to 95bf2eb Compare February 1, 2016 20:40
@efernandez efernandez force-pushed the change_covariance_model branch from 95bf2eb to 29835d8 Compare February 2, 2016 21:29
Both model also support a wheel resolution param, which is 0.0 by
default.
@efernandez efernandez force-pushed the change_covariance_model branch from aa240e2 to dc38d7f Compare February 8, 2016 15:56
@efernandez
Copy link
Member Author

Closing and creating a new PR on top of indigo-devel

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants