This is a work-in-progress (wip) repository at an attempt to write a DRL network for UAV attitude control.

The prior/ background work comes from this repo: https://github.com/PX4-Gazebo-Simulation
However more work needs to be done since most of the stuff involving mavros has not been updated since four years ago.

23 May 2022
Currently, in drl_blaster repo:
-> blasterDRL.py doesnt work as intended because:
--> Att_running msg is abit unstable. It requires the topic "att_running_msg" to be published which is now handled by an external offb_ctrl.py.
--> Action does not set/send anything.

TODO/TO FIX:

1. Try to incorporate an offboard function/node in the blasterDRL.py or create a separate script for putting the UAV in offboard.
2. To relook into actions, make sure to be able to send the required info (i.e. thrust etc.).
3. If attempting to redo, prepare for attitude/position (x,y,z) instead of just z.
4. How to get thrust states from mavros (FIXED with /setpoint_raw/target_attitude).
5. To check on env_ip and its publisher.

24 May 2022

1. Amended pub topic to sent to (Topic name)/mavros/setpoint_raw/attitude (Type)mavros_msgs/AttitudeTarget instead of 'input'.

25 May 2022
Added new offb ctrl script to keep UAV in offboard mode.
Note: script sends UAV into hold mode and land sequentially after receiving a "True" msg on /mission_status.

26 May 2022
Add an attiTarget.py to test if the UAV responds to external publishing of topics to (topic) /mavros/setpoint_raw/attitude when in offboard mode.
blasterDRL.py working now but..
TODO/TOFIX:

1. To check on saving to result_output.txt and h5 model. The model seems small. Not sure if its too easy so less data or..

27 May 2022
model not saving because new_trial and done not false at the same time,
TODO/TOFIX:

1. To check on line 298 to see if can rewrite that logic there, see if you can bring the model.save out instead of cascading it within.

24 June 2022
TODO/TOFIX:

1. To check on line 298 to see if can rewrite that logic there, see if you can bring the model.save out instead of cascading it within.
2. To fix line line 146 on IndexError --> Just implemented a quick and dirty way, need to revisit!
3. To figure out why the algo not sending commands to fight the disturbance from attiTarget.py