Connection issue

[elim17]

Based on what was said on this page https://github.com/USC-ACTLab/crazyswarm/issues/32 I should not have to run a vrpn client node to connect to motive Optitrack. When I do not, however, nothing shows up in Rviz. We are running Optitrack on its own workstation that has windows. On a separate workstation that runs linux, we are running the crazy swarm software. Any ideas on what is causing the disconnect when not using vrpn? Thank you!

[whoenig]

For Optitrack related issues, please verify that NatNet works for you, using the following repository: https://github.com/whoenig/NatNetSDKCrossplatform. The Crazyswarm internally simply uses that SDK for communication with OptiTrack.

[elim17]

Thank you. I ran the natnettest and mcr and both seem to have run property. I am still unable to see the crazyflie in Rviz without running the vrpn client. Is there anything else that could be causing this problem. Do I need to manually open NatNet before running hover.launch?

[whoenig]

No need to run anything else. The next step is to run mocap_helper.launch. In all cases, make sure that the launch file is configured correctly with the motion capture type and hostname/IP of the machine that runs Motive.

[elim17]

Ok. When I run mocap_helper.launch it seems to run alright. Does this output look alright to you?
When I run the hover_swarm.launch file I still do not see anything in Rviz.

15:41:28|~/crazyswarm$ source ros_ws/devel/setup.bash
removed '/tmp/setup.sh.OyDqWL8Qsa'
15:42:42|~/crazyswarm$ roslaunch crazyswarm mocap_helper.launch
... logging to /home/hcrobotics/.ros/log/c624d9bc-77ba-11eb-9523-40b0769f73bd/roslaunch-hcrobotics-6718.log
Checking log directory for disk usage. This may take a while.
Press Ctrl-C to interrupt
Done checking log file disk usage. Usage is <1GB.

started roslaunch server http://192.168.1.100:44147/

SUMMARY
========

PARAMETERS
 * /mocap_helper/motion_capture_type: optitrack
 * /mocap_helper/object_tracking_type: motionCapture
 * /mocap_helper/optitrack_local_ip: 192.168.1.100
 * /mocap_helper/optitrack_server_ip: 192.168.1.69
 * /rosdistro: melodic
 * /rosversion: 1.14.10

NODES
  /
    mocap_helper (crazyswarm/mocap_helper)

auto-starting new master
process[master]: started with pid [6728]
ROS_MASTER_URI=http://localhost:11311

setting /run_id to c624d9bc-77ba-11eb-9523-40b0769f73bd
process[rosout-1]: started with pid [6739]
started core service [/rosout]
process[mocap_helper-2]: started with pid [6746]

This is my mocap_helper.launch file if that helps. For Optitrack do we need the hostname and IP's or just the IP's?

<?xml version="1.0"?>
<launch>

  <node pkg="crazyswarm" type="mocap_helper" name="mocap_helper" output="screen" >
    <rosparam>
      # tracking
      motion_capture_type: "optitrack" # one of vicon,optitrack,qualisys,vrpn
      object_tracking_type: "motionCapture" # one of motionCapture,libobjecttracker
      #vicon_host_name: "vicon"
      #optitrack_host_name: "optitrack"
      # qualisys_host_name: "10.0.5.219"
      # qualisys_base_port: 22222
      #vrpn_host_name: "192.168.1.69"
      optitrack_local_ip: "192.168.1.100"
      optitrack_server_ip: "192.168.1.69"
    </rosparam>
  </node>
</launch>

[whoenig]

Just the optitrack_host_name is used in the current code. There should be a print out of markers and/or rigid bodies on the terminal if it is working correctly, similar to mcr.

[elim17]

@whoenig I changed the launch files to just use optitrack_host_name set to our Optitrack server IP address, but I still do not have any print out like that of mcr when I run mocap_helper.launch. Is there a flag I need to use to get the print out of markers?

[lkyfly0765]

Please make sure that when you run the ./mcr, the output contains the position information of your rigid body, and double check whether they are correct or not.

It seems weird that your ./mcr runs properly but have problems in Rviz. Please try to run the ./mcr program in another terminal window and then roslaunch your launch file, this should work to solve your Rviz issue.

Your mobcap_helper.launch is properly configured according to your information.

[whoenig]

OK - so it looks like it receives data, but it's not parsing it correctly. Which Motive version are you using? We only tested with 2.X. Also, some users reported that you have to only enable what you need in the streaming pane in motive. For example, if you only want to get the raw point cloud, only enable "unlabeled markers" in Motive to be streamed out.

[whoenig]

The "proper" fix will be to go over the parsing code in https://github.com/USC-ACTLab/libmotioncapture/blob/master/src/optitrack.cpp#L333-L630, and make sure it semantically matches the code in mcr. Unfortunately, all of this code is super ugly (not our fault, that's just how the OptiTrack SDK is...)

[elim17]

We are using Motive: Tracker 2.2.0. Could you explain more about raw point cloud vs rigid body? Our first goal is to be able to run hello_world.py and get one drone to hover.

[whoenig]

There is some documentation here: https://crazyswarm.readthedocs.io/en/latest/configuration.html#adjust-configuration-files. You essentially have to decide if you want Motive to track your object or libobjecttracker. I can only recall testing the latter with Optitrack myself. In either case, make sure that in the Motive Streaming pane you disabled the part of the stream that you do not want, because there seems to be some decoding bug in libmotioncapture that causes the crash.

[elim17]

Our lab decided to go with Motive to track our rigid bodies. Your point about parsing was correct. We were receiving too much data from Motive via NatNet. By turning off the point cloud data mocap_helper.launch was able to run correctly! NOTE: we did not need t have ./mcr running at the same time for it to work. Thank you for the help!