Skip to content

Deep Reinforcement Learning (RL) for target tracking with Autonomous Underwater Vehicles (AUV) using SB3 and HoloOcean simulator

License

Notifications You must be signed in to change notification settings

Ice-mao/RL_AUV_tracking

Repository files navigation

RL_AUV_tracking

A project for my major related graduation paper.

Using reinforcement learning(RL) to train an agent to tracking the target in the unknown underwater scenario in HoloOcean.

Quick Start

To run the simulation, first install all dependencies

  • HoloOcean==1.0.0
  • Stable Baseline3
  • pynput
  • bezier
  • filterpy
  • inekf
  • scipy
  • sb3-contrib
  • seaborn
  • shapely

if you want to run in my scenario,I give the scenario link below: https://drive.google.com/drive/folders/1MdT8NMozJARde7zL5kKebBi4WULfv2KC?usp=drive_link

you should copy the folder into /home/'yourname'/.local/share/holoocean/0.5.0/worlds/

Then simply run the script

python SB3_learning.py --env TargetTracking1 --map TestMap_AUV --nb_envs 5 --choice 0 --render 0 

choice:(0:train 1:keep training 2:eval)

render:(0:false 1:true)

Simulation Process

simulation

Additional information

If you want to know more details,you should read the code. :smile:

Or please keep staying tuning!

Something mentioned

Just for single target, mutitarget task needs revise the code

(revise the target0 -> target+str(rank))

About

Deep Reinforcement Learning (RL) for target tracking with Autonomous Underwater Vehicles (AUV) using SB3 and HoloOcean simulator

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages