Real-world maze navigation with tracking and experimental data collection.
Required scripts and files are mostly contained within this directory, which is to be located at ~/ExperimentPlatform
. Program expects ROS environment to be already set up at ~/catkin_ws
.
Both platforms have been set up to automatically log in and connect to the Belkin router on startup, with the network name powerbot
. The router reserves 192.168.0.2
for the older platform, and 192.168.0.3
for the newer platform. To control through VNC, use vnc://192.168.0.3::5900
in TightVNC Viewer on Windows. Headless access can be made through ssh [email protected]
from a lab workstation.
- In a terminal window, navigate to
ExperimentPlatform
and runmapping_launch.sh
cd ~/ExperimentPlatform
bash mapping_launch.sh
- Navigate the environment slowly with the main joystick, monitoring the mapping on Rviz
- Once done, run the following in a new terminal, replacing <map_name> with a file name of your choice
rosrun map_server map_saver -f <map_name>
For example, to save in the ExperimentPlatform
folder:
E.g. rosrun map_server map_saver -f /home/sinapse/ExperimentPlatform/new_map
- A .pgm and .yaml file will be generated, transfer them to the
ExperimentPlatform
folder if not already done so - Edit line 8 in
ros_launch.sh
to point to the new map OR rename new map to match the script
- In a terminal window, navigate to
ExperimentPlatform
and run 'ros_launch.sh`, which sets up the required ROS services.
cd ~/ExperimentPlatform
bash ros_launch.sh
- Start up the iOS app Runner on the iPhone.
- In another terminal window, navigate to
ExperimentPlatform
and runexp_launch.sh
to run required python scripts that control experiment behavior.
cd ~/ExperimentPlatform
bash exp_launch.sh
- In Rviz, get a good localization of the platform by using
P
to make a initial guess on position, and driving it around slightly - In the GUI window that pops up, Select the reward map under the
Reward Locations
drop-down list - To modify the reward map, use the
G
key in Rviz to redefine the 6 poster locations/directions (click and drag), with FIFO overriding -
- To create a new reward map, create a new csv file in the RewardData folder. In the GUI window, choose the new csv file from the drop-down list. Use the G key in Rviz to define the poster locations/direction (click and drag). The csv file will update the data accordingly.
- Load or modify other experiment settings, and save the settings if you wish
- Press
Start
Rviz will now display current goal. In the new window that pops up, you can monitor the experiment progress and intervene as required. To intervene, press P
to toggle Pause behavior. Note that the experiment will only pause at the end of the current trial. Once end of trial is reached with the Pause option, use R
to resume the experiment, or T
followed by Y
to terminate the entire experiment. Do NOT quit the experiment by any other method (eg. closing the window) as that will cause some ROS data to be recorded continuously. See below for remedy.
If the program exits ungracefully, open a new window and run
rosnode kill my_bag
rosnode kill my_bag2
At the end of all the trials, terminate the experiment with T
Y
. You can press Start
again in the GUI to re-run the whole experiment from Trial 1.
Experimental data will be stored in a folder named as the current date. A brief explanation of what each file contains:
- eparams.pkl - pickled python dictionary that stores parameters defined in the GUI
- platformPositions.csv - pose data timestamped with psychopy clock [time, dim1, dim2, sum(covariance)]
- sessionTriggers.csv - trigger values, timestamped with psychopy clock
- rospose.csv - raw pose data from ROS with ROS timing
- posebag_???-??-??-??-??-??.bag - rosbag data file, parent of rospose.csv
- rostrig.csv - trigger values, with ROS timings
- pose_bag???-??-??-??-??-??.bag - rosbag data file, parent of rostrig.csv
- save bag file with
rosbag record -a -o data.bag -x “/RosAria/*”
- playback the rosbag file
ros set use_sim_time true && rosrun rviz rviz -d rviz_config/default.rviz
# in another terminal
rosbag play <bag_file>.bag
- convert to csv file
rostopic echo <topic_name> -b <bag_file>.bag > <csv_file>.csv
- Eyelink Calibration is done via this
# in `powerbot` terminal
cd ~/EyelinkCalibration/EyelinkCalibration && python main_gui.py
- run calibrate_video.py after successful calibration for set 1 experiment
- super-position results can be found here. This is slightly different from Matlab version (using 'v4' griddata, wherease python's griddata does not have 'v4', hence failed for extrapolation)
Please refer to Developer's Guide for breakdown of how the python scripts interact. Possible features and suggestions will also be listed there.