Skip to content

This Repository is part of the European research Project SeaClear. This Project will include how an agent can make long horizon decisions to gain efficiency information to Map litter on the seabed with a camera and a Sonar and already given bathymetry data.

Notifications You must be signed in to change notification settings

Mateus224/DeepRL_3D-Liter_Mapping

Repository files navigation

To train the agent:

python3 main_keras.py --dueling true --train_dqn

To run the code with a trained agent:

python3 main_keras.py --dueling true --test_dqn_model_path saved_dqn_networks/DDDQN_Litter_Agent_2420000.h5 --test_dqn

To run the agent with visual explanation:

for visual explanations you need scipy=1.2.0

python3 main_keras.py --dueling true --test_dqn_model_path saved_dqn_networks/DDDQN_Litter_Agent_2420000.h5 --test_dqn --visualize --num_frames 300

About

This Repository is part of the European research Project SeaClear. This Project will include how an agent can make long horizon decisions to gain efficiency information to Map litter on the seabed with a camera and a Sonar and already given bathymetry data.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages