sunghoonhong / airsimdrl Goto Github PK

Dear Mr. Hong
Thanks for sharing us such a nice project :)
I have launch the code successfully, but the training process seems to be kinda slow. It would be so kind of you if you could share us your settings.json file for AirSim so that I can have an appropriate clockspeed or some other details.

업무 잘 되세요！

'Functional' object has no attribute '_make_predict_function',why?

I have some questions about your code.
what's the meanig of goals list (goals = [7, 17, 27.5, 45, goalY] ) and what's the level of a list in computing the reward ( goals[self.level] )?
I just know the fifth figure(57) is the goal position in y axis.
would you like to tell the other figures' meaning.
Thank you so much

Hello dear sunghoonhong,
thanks alot for the interesting code :)

I am training the agent via the rddpg_per algorithm and don't know exactly how long should I train it niether how to stop the training and run the trained model.

Can you please help me with that?
Thanks

Hello, how to run it

Hi,

Amazing project. Can you please share the Unreal Engine environments / projects? We would like to update this to AirSim 1.6.0 to use the Lidar sensors as well for coverage measurements.

I can't run your code

Hi, can the same model be also used in some other environment such as City / Building right off the bat?

Can you please share the requirement.txt file with us so that we can effectively execute this project?

There, I found this code is very interesting. But I do not have the environment used for training this.

Do you mind releasing the Airsim environment along with the code?

Thank you!!

Hongbo

Three scenarios will flash back when it runs for a period of time.

Hi, I'm using the --load-model and --play flags with tqd_per.py but it seems that the saved model is not used or does not perform well even on Easy, the quadrotor crashes straight ahead.

Tensorflow v1, Keras, AirSim 1.2.0 used

Can the same model be also used in some other environment such as City / Building right off the bat?

Hi,I'm running the td3_per.py,and I amended parser.add_argument('--play', action='store_true') to parser.add_argument('--play', action='store_true',default=False) for training.
But when the episodes increased by 100， the programs would stop and report the following error:
InvalidArgumentError (see above for traceback): You must feed a value for placeholder tensor 'batch_normalization_1/keras_learning_phase' with dtype bool [[Node: batch_normalization_1/keras_learning_phase = Placeholder[dtype=DT_BOOL, shape=[], _device="/job:localhost/replica:0/task:0/gpu:0"]()]] [[Node: dense_8/BiasAdd/_1409 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/cpu:0", send_device="/job:localhost/replica:0/task:0/gpu:0", send_device_incarnation=1, tensor_name="edge_125_dense_8/BiasAdd", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/cpu:0"]()]]
I don't know what caused this error，can you please help me with that?
Thank you！