| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
| |
also include the survive_config.h header to avoid various
implicit function declaration warnings on gcc
|
| |
|
|
|
|
|
| |
Added a raw pose output/ callback that the posers can
call when they have calculated a pose.
|
|
|
|
|
| |
The accelerometer will be used to determine "up" instead of
blindly using +z during calibration
|
|
|
|
|
|
|
|
|
| |
Works with 2 lighthouses.
Tracking from both lighthouses agree *much* better than before
Inverting the tracker no longer screws up tracking
Still much work to do to remove all axis angle and
speed up/ make predictable the algorithm to estimate the rotation of the
LH relative to the tracked object.
|
|
|
|
|
|
|
|
| |
Rotation was previously approximated using axis/angle
This change starts down the path of using quaternions exclusively.
This change appears to give at least as good as answers as the axis/angle
model in basic cases (also only tested with 1 lighthouse), but it is
currently much slower and runs in unpredictable time.
|
|
|
|
|
| |
Added some config options. Changed a few defaults. Better output from PoserTurveyTori.
Hopefully, this will make it easier for someone new to get up and running more easily.
|
| |
|
|
|
|
| |
This definitely works a little better. There are still issues.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
Also updated the visualization in calibrate.c to show
position & orientation as calculated by each lighthouse
at the same time
|
|
|
|
|
|
|
| |
This feature removes the requirement that you calibrate
with the tracked object in any particular orientation.
The IMU on the tracked object will be used to determine
what direction is "up"
|
|
|
|
|
|
| |
quattomatrix33() needed to be updated after the fix in quatfrom2vectors
Additional poser updates to make it more robust.
Poser is now updating based on both lighthouses. Expect significant jitter from this.
|
| |
|
|
|
| |
I've been seeing starvation that manifests itself as weird light data coming into the disambiguator. A major cause ended up being starvation. This change will cut the frequency of poser calls by by 75%
|
| |
|
| |
|
|
|
|
|
| |
Haven't tested it yet, low confidence it's right,
but the code is on the right track.
|
| |
|
|
|
|
|
|
|
|
| |
The z=0 plane is now initially aligned to the
device's z=0 plane during calibration.
(works really well for the tracker, which is
oriented with Z in the proper direction when
sitting on a horizontal surface)
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
Sometimes the search algorithm would spend way too much time making a bunch of small improvements, when it would have been better to just stop and call it "good enough"
|
| |
|
|
|
|
| |
There's a ton of code cruft, and the algorithm is currently too slow. BUT I can track an object using only 1 lighthouse for tracking, at (I believe) an update rate of at least 7.5 HZ. By tracking, I know the position and orientation of the lighthouses relative to the tracked object, and I know the tracked object's location relative to the lighthouse. I don't have the orientation of the tracked object relative to the lighthouse yet, but that should be easy given the rest of the "knowns."
|
| |
|
|\
| |
| |
| |
| | |
# Conflicts:
# src/poser_turveytori.c
|
| | |
|
|/ |
|
|
|
|
| |
Lots of cruft. Lots of room for improving performance.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
OctavioRadii converges to the right radius!!!!
|
|
|