Package for simulating wheel loader in gazebo engine

This page provides instructions and source code for simulating wheel loader and a pile of particles in gazebo. This is complementary material to the paper

Wheel Loader Scooping Controller using Deep Reinforcement learning submitted to the IEEE Access (2020).

We provide a platform in which the user can independently collect data on a simulator using the basic envs in environments folder. Within /src/environments/komodo_env.py the user can change pile features (shape, amount, etc..) which defined in the Pile class. The user can also use /src/environments/komodo_env_new.py for diffrent pile arrangements on every episode (currently support up to 3 diffrent arrangements)

Files

komodo2/launch/komodo2.launch : (parameters: depth_cam,lidar,gazebo): lauch all the required nodes.

komodo2_control/config/komodo2_control.yaml : joint controller config for Gazebo

komodo2_description/urdf/komodo2.xacro : urdf model in real robot

komodo2_description/urdf/komodo2_gazebo.xacro : urdf model in Gazebo

komodo2_rl/src/agents/ddpg.py : Deep Deterministic Policy Gradient agent (tensorflow)

komodo2_rl/src/agents/a2c.py : Advantage actor-critic A2C agent (tensorflow)

komodo2_rl/src/agents/utils.py : utilities (replay buffer, OUNoise

komodo2_rl/src/environments/Spawner.py : object spawner script (can be edit to include other shapes)

komodo2_rl/src/environments/komodo_env.py : environment setup in Gazebo

komodo2_rl/src/environments/komodo_env_new.py : environment setup in Gazebo with random piles (3)

komodo2_rl/src/komodo_learn.py : learning script ( change param ddpg/a2c)

komodo2_rl/src/komodo_gazebo_act.py : learned agent acts in Gazebo

komodo2_rl/src/komodo_model.py : enviorment setup real robot

komodo2_rl/src/komodo_act.py : learned agent takes actions in a real robot

komodo2_rl/src/komodo_maunal_control.py : manual control node using joy package

Prerequisites

• python 2.7
• ROS Kinetic Kame
• Tensorflow >= 1.4
• Gazebo simulation

additional installations for simulation

• ROS control

sudo apt-get install ros-kinetic-ros-control
sudo apt-get install ros-kinetic-ros-controllers
sudo apt-get install ros-kinetic-gazebo-ros-control

How to

Training in Gazebo

roslaunch komodo2 komodo2.launch gazebo:=true
python komodo_learn.py

Learned model in gazebo

roslaunch komodo2 komodo2.launch gazebo:=true
python komodo_gazebo_act.py

learned agent in real robot

roslaunch arm_control arm.launch
rosrun arm_control inv_control.launch
roslaunch komodo2 komodo2.launch lidar:=true
python komodo_act.py