RL_code

Model based RL with ENN The results plots in images folder, with videos of the training process.
The code in the mpc.py file, environments in env folder.

V-REP simulator (CoppeliaSim) is used, and needed to install.

wget http://coppeliarobotics.com/files/V-REP_PRO_EDU_V3_5_0_Linux.tar.gz
tar -xf V-REP_PRO_EDU_V3_5_0_Linux.tar.gz
rm V-REP_PRO_EDU_V3_5_0_Linux.tar.gz

(Experiments with Rozum in VREP 3.5, but KUKA in CoppeliaSim-latest version)

If it is running slow try install vrtualgl

curl -O -J -L https://sourceforge.net/projects/virtualgl/files/2.6.3/virtualgl_2.6.3_amd64.deb/download
dpkg -i virtualgl_2.6.3_amd64.deb
apt install -f
/opt/VirtualGL/bin/vglserver_config -config +s +f -t
rm virtualgl_2.6.3_amd64.deb

After that run

rmmod nvidia
# if : rmmod: ERROR: Module nvidia is in use by: nvidia_uvm nvidia_modeset
modprobe -r nvidia