If you added the privisioning script to your
~/.bashrc, you can directly execute the algorithm. Load the environment variables manually otherwise.
cd ~/gym-gazebo2/examples/MARA python3 gg_random.py -g
Every MARA environment provides three command-line customization arguments. You can read the details by using the
-h option in any MARA-script (e.g:
python3 gazebo_mara_4actions.py -h). The help message is the following:
usage: gg_random.py [-h] [-g] [-r] [-v VELOCITY] [-m | -p PORT] MARA environment argument provider. optional arguments: -h, --help show this help message and exit -g, --gzclient Run user interface. -r, --real_speed Execute the simulation in real speed and using the running specific driver. -v VELOCITY, --velocity VELOCITY Set servo motor velocity. Keep < 1.57 for real speed. Applies only with -r --real_speed option. -m, --multi_instance Provide network segmentation to allow multiple instances. -p PORT, --port PORT Provide exact port to the network segmentation to allow multiple instances.
If you want to get faster simulation speeds, you should launch the simulation withouht the visual interface of gazebo. But it is possible that you want to check the learnt policy at some point without stoping the training, so this is how you do it:
-m --multi_instanceoption to provide network segmentation, do the following:
In a new terminal, set the corresponding
GAZEBO_MASTER_URI: For convinience, this environment variable is printed at the beginning of every Env execution. Just copy and export it. You can also find information related to any running execution inside
/tmp/gym-gazebo2/running/ folder. Example:
Finally launch the client:
Final note: you can launch as many
gzclient instances as you want as long as you manage properly the
GAZEBO_MASTER_URI environment variable.