-
Notifications
You must be signed in to change notification settings - Fork 125
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
JPDA and SystematicResampler Fixes (updated) #193
Conversation
c74d2bc
to
f431b5e
Compare
Codecov Report
@@ Coverage Diff @@
## master #193 +/- ##
=======================================
Coverage 92.47% 92.48%
=======================================
Files 113 113
Lines 4027 4030 +3
Branches 576 577 +1
=======================================
+ Hits 3724 3727 +3
Misses 223 223
Partials 80 80
Continue to review full report at Codecov.
|
@sglvladi - Could you please rebase this to drop the delete commits, and squash duplicate commits. |
Particles are sorted before calculating the cumulative sum array. Thus when particles are resampled, they are drawn from the ordered distribution, rather than randomly.
The prediction and measurement prediction were originally set for every hypothesis in the multihypothesis as for the missed detection state. This causes spurious results at the update step if the measurements in the multihypothesis are asynchronous (e.g. grouped by a time-windowing feeder class), and the sensor platform is moving. In this case, the updated state is calculated using the sensor's translation offset at the missed detection timestamp (i.e. the latest timestamp in the window), but for a position measurement taken at the measurement timestamp (which may be earlier in the window). This fix attaches the prediction/measurement prediction calculated for the measurement to the relevant SingleHypothesis, such that the update is calculated using the correct sensor offset for the measurement timestamp.
@sdhiscocks I've had a go at it. Let me know if there's anything else that needs doing. |
Thanks @sglvladi |
This PR supersedes #152. The following changes have been applied:
master
commit.attrgetter
is now used when sorting the weights inSystematicResampler
JPDA
.