evs

This package provides a ROS node that generate an assumed reference camera's trajectory based on one obtained using a visual servoing scheme. The resulting data can be then handled, evaluated and compared using the evo tools (https://github.com/MichaelGrupp/evo).

Expected input data:

A ROS bag that contain, at least, two "geometry_msgs/PoseStamped" topics, one for the camera pose at each iteration of the visual servoing, and one for the desired camera pose (see the sample code below to create a ROS bag in cpp).

rosbag::Bag vsbag;
vsbag.open("vs.bag",rosbag::bagmode::Write);
[VS INIT]
vsbag.write("/vs/camera/desired_pose", ros::Time::now(), desiredRobotPoseStamped);
[VS MAIN LOOP START]
    vsbag.write("/vs/camera/current_pose", ros::Time::now(), currentRobotPoseStamped);
[VS MAIN LOOP END]
vsbag.close();

Usage:

• Edit the launch file parameters:
  • pathToInputBag: path to find the VS ROS bag (see the sample code above)
  • desiredPoseTopic: name of the desired pose topic in the input ROS bag (e.g., "/vs/camera/desired_pose")
  • currentPoseTopic: name of the current poses topic in the input ROS bag (e.g., "/vs/camera/current_pose")
  • pathToOutputBag: path to save the generated ROS bag data usable with evo
• Then, roslaunch generateVSReferenceTrajectory.launch

The resulting ROS bag can be used with evo tools:

• evo_traj bag output1.bag /vs/reference /vs/current -p
• evo_ape bag output1.bag /vs/reference /vs/current --save_results output1.ape
• evo_ape bag output2.bag /vs/reference /vs/current --save_results output2.ape
• evo_res output1.ape output2.ape -p --use_filenames
• etc... (for more, see https://github.com/MichaelGrupp/evo)