PorkPy
commented
5 years ago So I found this 'quick start guide' https://gist.github.com/danielsnider/e09c56a96b187eec23c482f9400e59a1 which showed me where I needed to input my robot's IP address. And now it finally moving! Heres the guid for UR5 just incase the page goes all 404 on you!
SenseAct Quick Start This guide has been tested on Ubuntu 16.04.
Install General Dependecies sudo apt update sudo apt install python3 python3-dev python3-pip python3-tk cmake libopenmpi-dev zlib1g-dev git Install MuJuCo Robotics Simulator The SenseAct learning task in simulation uses a popular physics simulator called MuJoCo. First install the prerequisites for MuJoCo:
sudo apt install patchelf libgl1-mesa-dev libgl1-mesa-glx libglew-dev libosmesa6-dev Then install MuJoCo according to their install guide.
Sign up for a 30-day free trial on the MuJoCo website or free license if you are a student. The license key will arrive in an email with your username and password. Download the MuJoCo version 1.50 binaries Linux, OSX, or Windows. Unzip the downloaded mjpro150 directory into ~/.mujoco/mjpro150, and place your license key (the mjkey.txt file from your email) at ~/.mujoco/mjkey.txt. Add MuJoCo to your PATH:
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/dan/.mujoco/mjpro150/bin echo "export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home/dan/.mujoco/mjpro150/bin" >> ~/.bashrc Install SenseAct Download SenseAct:
git clone https://github.com/kindredresearch/SenseAct cd SenseAct (Optionally) Create a python virtual environment:
pip3 install virtualenv virtualenv venv source venv/bin/activate Install SenseAct's python dependencies:
pip3 install -e . pip3 install baselines==0.1.5 Run SenseAct For your first learning task, we'll simulate the robotic control of a double inverted pendulum in a MuJuCo simulation.
Enable visualization of the simulated world in MuJoCo:
sed -i 's/is_render=False/is_render=True/g' ./examples/advanced/sim_double_pendulum.py Start learning:
python3 ./examples/advanced/sim_double_pendulum.py Hardware Setup UR Robot Arm Tested on UR Software v. 3.3.4.310. To control a UR robot arm you will need its IP address and be connected to the same network. You can find the IP address on the teach pendent under network settings.
Once you have the IP address of the arm enter it here:
SenseAct/examples/advanced/ur5_reacher.py
# Create UR5 Reacher2D environment
env = ReacherEnv(
setup="UR5_default",
host="192.168.5.20", <-- Set IP hereNow start learning:
python3 examples/advanced/ur5_reacher.py If you want to use a pretrained model:
git checkout load_model python3 examples/advanced/ur5_reacher.py ./examples/advanced/pre_trained_models/ur_reacher_2_trpo.pkl
I'm happy I've finally found an implementation of reinforcement learning for universal robots. I've had an ongoing issue with trying to interface my learning algorithm with a real robot. I've looked at rllab with ros but couldn't get anything to work. I don't know enough to rewrite an algorithm in a ros node. I've even looked at using opcua to try and import/export data between ros and my learning algorithm.
I've installed SenseAct ok and can run the example inverted pendulum. However, I can't seem to find any documentation on how to implement anything on a real robot. I have a ur5 and I'm using ubuntu16.04. Where is the communication between the ur5 and the algorithm? How do I initiate this communication? Is there a tutorial on building an environment with a real ur5 and testing different algorithms? I'm 18 months into a 3-year phd on robotics and reinforcement learning and I still haven't figured out how to interface between some algorithm and my robot. Please help.