Stability observer #88
Open
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
stefanscherzinger
changed the title
stefanscherzinger
force-pushed
the
add-stability-observer
branch
from
776b4e5 to
076573b
Compare
stefanscherzinger
force-pushed
the
add-stability-observer
branch
from
076573b to
2cc243f
Compare
The first idea is to use a prototype implementation in Python with good support for FFT computation in numpy. We currently listen on force-torque sensor signals and test different computations on the incoming data.
This makes sure that our FFT computations use the sensor rate that we actually achieve with our performance. Also clarify some variables.
That's now robust when the sensor values contain only zeros.
It's stable now for sensor callbacks with an arbitrary rate.
We need to drop the zero-frequency term. According to the `numpy` documentation, the zero-frequency term is the *sum of the signal*. That's not what we want. The range of frequencies and the amplitudes now look meaningful for test data.
This is the latest state I was working on. I remember that things were not as straight forward as expected and that I questioned whether the paper's theory could actually work on my system.
stefanscherzinger
force-pushed
the
add-stability-observer
branch
from
c08ebee to
6327fcd
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Motivation
The stability of the
cartesian_force_controllerand thecartesian_compliance_controllerdepends on the environment's stiffness and varies with each use case. While feeling natural and responsive in free motion, the robot's sensitivity normally is decreased in contact to remedy oscillations. Users thus often set the controller gains conservatively low for the worst case, i.e. loosing some of the possible sensitivity in free motion.Approach
Frequency domain-based stability observers could automatically detect oscillations (before they become apparent to the user's eye and touch) and decrease the controller gains adequately. Here's a very promising approach that does this directly on the force-torque sensor signals. Let's see if we can integrate this into the
cartesian_controllers.Steps
error_scale. Another alternative is increasing the joint damping. Or link_mass.