Hello, I would like to conduct parallel testing in multiple environments, but I cannot find where to modify the IP address. Can you please tell me how to operate so that I can run multiple dogfight windows with different IP addresses on a single computer?

How can the Dogfight software interface with OpenAI Gym library? Can you make a demo, I think it's very useful for learning and the application of your platform.

Hello, how is it compatible with ubuntu18.04 , because it needs to be upgraded to GLIBC_ 2.29, this is a very dangerous operation. Can you help me?

What does the position and rotation of the aircraft refer to?
I can currently obtain the quantitative information of the speed of the aircraft, but cannot obtain the speed vector information of the aircraft. How can I obtain it?

When performing reinforcement learning on an aircraft, it is common to turn on renderless mode to speed up training, which is the following code:
df.set_renderless_mode(True)
May I ask if the following code can be used to turn off the client update mode during reinforcement learning training?df.set_client_update_mode(False)
Does this affect my training results?

Dear author：
I’m interested in how does the IA mode work？I'm curious about how did you train the agent to become “intelligent agent”？Which algorithm did you use？

when i run python main.py , running to hg.RenderInit(Main.win, hg.RT_OpenGL), it give me an Error:

Because I want to train them on the server, but both the local machine and the server are ubuntu18.04, so I created a container of ubuntu20.04 with the docker, and installed harfang and doglight in it. But because the docker and the server can not be displayed graphically, the error occurred in running main.py: window_init error. And, it cannot work when I set "OpenGL": true in config.json. Therefore, how can I make it work without graphical display, just like StarCraft

Where is the code related to "autopilot" ?

Hello, I would like to ask if the environment can only support up to 4 drones? Is there a function of adding fighter jets to study the application of multi-agent reinforcement learning in air combat？

When I run the main.py file it gives the following error:

Compiling assets...
0%|          | 0/100 [00:00<?, ?it/s]Traceback (most recent call last):
  File "/home/kadir/Desktop/kadirnar/github/dogfight-sandbox-hg2/source/main.py", line 32, in <module>
    dc.run_command("..\\bin\\assetc\\assetc.exe assets -quiet -progress")
  File "/home/kadir/Desktop/kadirnar/github/dogfight-sandbox-hg2/source/data_converter.py", line 79, in run_command
    for line in execute_com(exe):
  File "/home/kadir/Desktop/kadirnar/github/dogfight-sandbox-hg2/source/data_converter.py", line 74, in execute_com
    p = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
  File "/usr/lib/python3.8/subprocess.py", line 858, in __init__
    self._execute_child(args, executable, preexec_fn, close_fds,
  File "/usr/lib/python3.8/subprocess.py", line 1704, in _execute_child
    raise child_exception_type(errno_num, err_msg, err_filename)
FileNotFoundError: [Errno 2] No such file or directory: '..\\bin\\assetc\\assetc.exe assets -quiet -progress'
  0%|          | 0/100 [00:00<?, ?it/s]

What could be the cause of this error?

I apologize for my unfamiliarity with GitHub, as I accidentally deleted the previous issue. Allow me to restate the problem I encountered and the temporary solution:

While configuring dogfight on the new workstation, I encountered an issue. When the target device locks onto the opponent and the missile device launches the missiles, they consistently deviate from the target and continue to circle around it, even when the opponent is flying at a constant speed without any turns. However, when I configured it on my desktop computer and personal laptop, it worked properly. You can find a video showcasing the issue below:

My temporary solution was to modify the "angular_frictions" value in the missile parameters
https://github.com/ErkMkd/dogfight-sandbox-hg2/blob/69815f4804b87bf0ca250d2f49d371c6c9fc4dbd/source/missile_aim_sl.py#LL21C7-L21C7
and, as a result, I had to increase the value of "target_collision_test_distance_max"
https://github.com/ErkMkd/dogfight-sandbox-hg2/blob/69815f4804b87bf0ca250d2f49d371c6c9fc4dbd/source/Machines.py#LL876C54-L876C54
to resolve the issue.
Thanks for the previous responses from @astrofra and @ErkMkd .

Hi

I'm doing research about dogfighting between planes. I want to design some simple instances, such as a flare countermeasure, and add those to dogfight 2 environments. I wonder whether there is any manual of guidance. Thank you!

Hello,

I encountered an issue while configuring the dogfight on the new workstation. When the target device locks onto the opponent and the missile device launches a missile, the missile consistently deviates from the target, continuously circling around it, even when the opponent is flying at a constant speed without any turns. However, when I configured it on my personal laptop, it worked fine.

I would like to inquire about the possible solutions to resolve this issue. Could it be related to the hardware configuration? The GPU on my device is Nvidia GeForce RTX 3090.

Thank you for your assistance in advance.

Hi!

First of all, thank you for your great effort in the field of dogfight simulations and build this wonderful platform.

I have problems using "Network mode" with client_*.py code. I made a screen record, but the video is too big to be uploaded on Github, so I upload it on OneDrive. You could download it in the link below.

https://1drv.ms/u/s!AjYxuXuKCgSEvjGBt1HkUVXPusP-?e=cRpGYp

In this trial, the rendering couldn't be played correctly while the program and even the sound works well. When I tried to run client_sample.py, the rendering will play after the program is finished. So did the client_sample_missile_custom_phisics.py.

Besides, just a small suggestion. The IP and port displayed in the top-left corner of the screen would be hidden if the sandbox windows was in different size, along with health, bullets, etc.

I'm using Windows 10 Home edition,
CPU: Intel Core i7 11700
RAM: DDR4 32GBytes
GPU: NVIDIA GeForce RTX 3070

Can we integrate open-source dynamics like JSBSim to Harfang flight models. And also how to include support for HOTAS, rudder and throttle too.

Hello, is there some frameworks or flow charts for dogfight, can you provide it please?

if self.plane_heading>180:
deger_1 = (self.plane_heading-360)
else:
deger_1 = self.plane_heading

    if self.plane_heading_2>180:
        deger_2 = (self.plane_heading_2-360)
    else:
        deger_2 = self.plane_heading_2
    self.reward -= abs(deger_1-deger_2)/90

I don't know what's the meaning of this reward setting.
I don't know what's the meaning of this code.
What do these equations mean？
Thank you very much for your reply！

I've just had problems reading the code and am having trouble understanding the following parameters：
1.Aircraft.v_move and get_planestate(plane_id)["move_vector"]
2.Plane_Euler= [plane_state["Euler_angles"][0] / NormStates["Plane_Euler_angles"],
plane_state["Euler_angles"][1] / NormStates["Plane_Euler_angles"],
plane_state["Euler_angles"][2] / NormStates["Plane_Euler_angles"]]
3.Plane_Pitch_Att = plane_state["pitch_attitude"] / NormStates["Plane_pitch_attitude"]
Plane_Roll_Att = plane_state["roll_attitude"] / NormStates["Plane_roll_attitude"]
Plane_Hor_Speed = plane_state["horizontal_speed"] / NormStates["Plane_horizontal_speed"]

Run exe output:
`(0:051:583:200) Harfang core 1.2.0 (build 1538b3acd3c9351b7b9108742a5f352c98b78e5d Feb 7 2019 09:51:22)
(0:056:555:200) See http://harfang3d.com/license for licensing terms
(0:204:585:600) Load plugins from 4 default paths
(0:206:580:300) Scanning for plugin in 'c:/Users/User/'...
(0:211:050:000) error: LoadSharedLibrary('sfextra.dll') failed, reason: %1 is not a valid Win32 application.

(0:215:589:600) 0 plugin(s) loaded
(0:217:320:100) Scanning for plugin in 'c:/Users/User/AppData/Local/Temp/_MEI184562/base_library.zip/harfang/'...
(0:222:930:200) 0 plugin(s) loaded
(0:224:276:900) Scanning for plugin in 'c:/Users/User/AppData/Local/Temp/_MEI184562/lib-dynload/harfang/'...
(0:229:429:900) 0 plugin(s) loaded
(0:231:235:200) Scanning for plugin in 'c:/Users/User/AppData/Local/Temp/_MEI184562/harfang/'...
(0:236:044:700) 0 plugin(s) loaded
(0:238:165:200) 0 plugin(s) loaded
(0:240:725:100) Mount root driver 000002236AEAE4B0 (relative to '')
(0:256:356:600) warn: No renderer available, was LoadPlugins called successfully?
Traceback (most recent call last):
File "flight_sim.py", line 4232, in
File "flight_sim.py", line 2575, in request_screen_mode
RuntimeError: RenderInit failed, was LoadPlugins called succesfully?`

I'm sorry to bother you， but I currently want to use this dogfighting software as a simulation environment to verify the reliability of some flight algorithms. How should I control the aircraft through flight data and what data do I need to control？

Hi! I wonder whether there is any paper about Dogfight 2 so that I could cite if I make use of this sandbox in my research? Or I can just cite this Github page? Thank you

Just download the source code for 1.07, How to directly run the game with python?

Thanks !

Hi！
I had some problems reading your citation "HARFANG3D DOG-FIGHT SANDBOX: A REINFORCEMENT LEARNING RESEARCH PLATFORM FOR THE CUSTOMIZED CONTROL TASKS OF FIGHTER AIRCRAFTS". I contacted you by email and did not get a reply.
Due to a lack of expertise and experience, we could not achieve the results shown in your paper. I would like to ask if I can get the source code and algorithm for the experimental part of the article
l am looking forward to your reply as soon as possible!

I have been looking for a Reinforcement Learning research platform for air combat maneuver generation for a long time, and it is very lucky to find your paper and platform. The platform is really amazing, and I will study your paper and projects carefully. I believe that I can do a lot of work through the platform.

When the combat is done, I want to see the combat process, so I think it would be better if there is a replay of the combat with flight route video playback.

Hi,
I am working on a reinforcement learning project using the Harfang platform and I would like to ask how to enable multiple dogfight-sandboxes in network mode to achieve parallel training?

As reported by Kadirnar the image appears upside down when running on Linux in OpenGL.

Add command's shortcuts in the README file

When I pip install harfang-.....-linux_x86_64.whl (import harfang is ok )and then run python main.py in the dogfight.../source, there is a error as follow:Compiling assets...
99%|█████████████████████████████████████████▌| 99/100 [25:30<00:15, 15.46s/it]
Assertion failed (ret == GLFW_TRUE)
Reason: Failed to initialize window system
RuntimeError: X11: The DISPLAY environment variable is missing

The python environment I'm using is the anaconda virtual environment, and the following error occurs when I open dogfight：
If I open it through the source main file it shows:
Harfang 3.2.4 for CPython 3.2+ on windows-x64 (build 2002087e407e25dcd76a597f4b180bb6c82a5332 Sep 19 2022 14:31:06)
See https://www.harfang3d.com/license for licensing terms
Compiling assets...
0%| | 0/100 [00:00<?, ?it/s]Traceback (most recent call last):
File "D:\Users\14290\Desktop\dogfight-sandbox-hg2-main\source\main.py", line 66, in
dc.run_command("../bin/assetc/assetc assets -api GL -quiet -progress")
File "D:\Users\14290\Desktop\dogfight-sandbox-hg2-main\source\data_converter.py", line 79, in run_command
for line in execute_com(exe):
File "D:\Users\14290\Desktop\dogfight-sandbox-hg2-main\source\data_converter.py", line 74, in execute_com
p = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
File "D:\ANACONDA\envs\harfang2\lib\subprocess.py", line 947, in init
self._execute_child(args, executable, preexec_fn, close_fds,
File "D:\ANACONDA\envs\harfang2\lib\subprocess.py", line 1416, in _execute_child
hp, ht, pid, tid = _winapi.CreateProcess(executable, args,
FileNotFoundError: [WinError 2] The system cannot find the specified file。
0%| | 0/100 [00:00<?, ?it/s]

If I open it through the start.bat it shows:
d:\Users\14290\Desktop\dogfight-sandbox-hg2-main>cd source

d:\Users\14290\Desktop\dogfight-sandbox-hg2-main\source>..\bin\python\python.exe main.py
'..\bin\python\python.exe' Not an internal or external command, not a runnable program
or a batch file

d:\Users\14290\Desktop\dogfight-sandbox-hg2-main\source>pause

hope you have a good day, and thanks for your great package
when i'm trying to run the game in python 3.10.3 and harfang newest version i get this error , is there anyone who knows about the problem?

e:\dogfight-sandbox-hg2-main\dogfight-sandbox-hg2-main\source\Machines.py:135: Warning: Invalid node
  nm = nd.GetName()

and another error when loading fps_start_point , i changed nothing by the way

In network mode, I'm trying to get two planes to start a dogfight and I can attack the enemy plane by doing "df.activate_IA", but how do I do it to get my plane to attack? It seems to only fly in circles

Multi-process and multiple doglight simulation starts are not feasible, because even if multiple dogEnv instances are created, they all import the same client, that is, the data received and sent from the same doglight simulation platform. How to realize multi-process？

I'm getting an error when using your latest version of network_client_example.
The terminal printout shows that my client's socket received a new error while decoding. The client's socket received the None which cannot be decoded.I would like to know the reason for this problem. Thank for your helping.

What is the difference between flag_destroyed and wreck?

VR doesn't seem to work after v1.0.2 , issuing this error (in the latest build):

G:\works\harfang\hg2\dogfight-sandbox-hg2\source\MachineDevice.py:130: Warning: Invalid scene animation
  self.gear_anim_play = self.scene.PlayAnim(self.open_anim, hg.ALM_Once, hg.E_Linear, hg.time_from_sec_f(self.gear_moving_delay), hg.time_from_sec_f(self.gear_moving_delay), True, 1)
Traceback (most recent call last):
  File "G:\works\harfang\hg2\dogfight-sandbox-hg2\source\main.py", line 154, in <module>
    Main.update()
  File "G:\works\harfang\hg2\dogfight-sandbox-hg2\source\master.py", line 1557, in update
    cls.render_frame_vr()
  File "G:\works\harfang\hg2\dogfight-sandbox-hg2\source\master.py", line 1204, in render_frame_vr
    vid = cls.sea_render.render_vr(vid, cls.vr_state, vs_left, vs_right, output_fb_left, output_fb_right) #, tex_reflect_left_color, tex_reflect_left_depth, tex_reflect_right_color, tex_reflect_right_depth)
  File "G:\works\harfang\hg2\dogfight-sandbox-hg2\source\planet_render.py", line 374, in render_vr
    focal_distance_left = hg.ExtractZoomFactorFromProjectionMatrix(vs_left.proj)
RuntimeError: incorrect number of arguments to function ExtractZoomFactorFromProjectionMatrix

I want to add an image to a region of the map. How can I do that?

Hello, I am sorry that I am not a major in aerospace and know little about aerodynamics of aircraft, so it is difficult for me to understand the kinematics model of aircraft in your code, such as the ‘matrix’ of update_physics function. Could you please give me some relevant reference materials to help me understand the specific aircraft kinematic model?

Hello, when we tested the IA mode of the aircraft, we found that when we were away from the enemy aircraft for a certain distance, the enemy aircraft did not seem to come close to us, just like the IA mode failed. What can I do to solve this problem?

Hello, I am currently conducting scientific research and development based on your software. I would like to ask what algorithm is the flight combat strategy of the enemy aircraft in this software based on?

May I ask where I can see the kinematic equations of the aircraft, that is, the equations for updating parameters such as position and speed. I have been searching for a long time in the code but have not found it.

When I set renderless_mode to ture, the training process is stopped by the Agent reporting an error for receiving an empty packet.

I want to run this library on ubuntu. Is there Ubuntu support?

Error Message:

from .harfang import *
ImportError: libpython3.6m.so.1.0: cannot open shared object file: No such file or directory

Hey,I'm interested in Multiagent Dogfight,I'd appreciate it if you can share me the code of the multiagent-agent reinforcement learning.

Hello, there is a reset error as follow:
socket_lib.py", line 91, in send_message sock.sendall(sizeb + message)
ConnectionResetError: [Errno 104] Connection reset by peer

Can you help me?

I'm a novice and interested in harfang doglight. I just read your paper.
Could you please share the code of the experiment part in your paper？Thank you so much!

Description: Giving the autopilot "0" heading command, the position obtained by the "get_plane_state" function is abnormal. Theoretically, the position "y" should increase and "x" remains unchanged (or small change). But the result is they are both changed after a period of time. As shown:

Thanks for your help!