A Python3, ROS-integrated, PyBullet-based Robot Simulator Foundation.

Installation Instructions

Install Python3 libraries for ROS (optional):

If you want to use Cairo Simulator with ROS, you need to enable Python3 ROS development and source the ROS environment scripts.

Using ROS is optional and the below steps are not strictly necessary.

sudo apt-get install python3-pip python3-yaml
pip3 install --user rospkg catkin_pkg
source /opt/ros/melodic/setup.bash

Install the dependencies:

If using ROS:

After sourcing the ROS environment scripts: pip3 install --user -r requirements.txt

If using a Python virtual environment (i.e. virtualenv, pyenv etc,.):

pip3 install -r requirements.txt

and then from the top level cairo-simulator directory:

pip3 install -e ./

Note that for some versions of Python, the igraph install will need to be compiled. You might have to install the below system dependencies:

sudo apt install build-essential python-dev libxml2 libxml2-dev zlib1g-dev bison flex automake autoconf

Installing cairo-planning-core / planning-core-rust

Clone the repo: https://github.com/cairo-robotics/planning_core_rust

This repostiory is installable as a Python package, however you will need the rust compiler installed on your computer: https://www.rust-lang.org/tools/install

With you virtual environment activated, run:

pip3 install <path_to_planning_core_rust_repo>

Running the Simulator

Unlike Gazebo, this is not a standalone executable program but rather is a foundation upon which a simulator can be created. The sawyer_test.py file contains a fairly minimal example showing how to use this package, initializing the simulator, a robot, and an object in the world.

Simulator Components

The simulator uses 3 primary components:

Simulator: A singleton class that wraps the PyBullet physics simulation environment and keeps track of all simulation objects.
Robot: A class that exposes common functionality relevant to robot control, including setting up ROS topics for joint position control and for broadcasting state.
SimObject: A class that facilitates loading passive objects into the simulation, with functions for managing position and orientation.

Robot Types

The initial version of this simulator only includes support for Manipulator arms, created by extending the Manipulator class. (See the Sawyer class in Manipulators.py for an example)

Manipulator

Manipulator arms expose functionality to:

publish_state (typically the configuration vector for all joint/gripper positions)
get_current_joint_states which returns a configuration vector
move_to_joint_pos which commands the arm to a given configuration
move_to_joint_pos_with_vel which moves to a target configuration with per-joint velocity specifications
check_if_at_position which returns whether the configuration of the manipulator is within some epsilon of a target configuration
execute_trajectory which moves the arm through a sequence of configuration waypoints at specific timestamps.
solve_inverse_kinematics which returns a configuration (list of joint positions) given a target position (and optional orientation) for the robot's end effector.

cairo-robotics/cairo_simulator