qibullet 1.0.2

Bullet-based simulation for SoftBank Robotics' robots

qibullet

Bullet-based python simulation for SoftBank Robotics' robots.

Installation

The following modules are required:

• numpy
• pybullet

The qibullet module can be installed via pip, for python 2.7 and python 3:

pip install --user qibullet

Usage

Please note that only the Pepper robot is currently handled by this module. A robot can be spawn via the SimulationManager class:

from qibullet import SimulationManager

if __name__ == "__main__":
    simulation_manager = SimulationManager()

    # Launch a simulation instances, with using a graphical interface.
    # Please note that only one graphical interface can be launched at a time
    client_id = simulation_manager.launchSimulation(gui=True)

    # Spawning a virtual Pepper robot, at the origin of the WORLD frame
    pepper = simulation_manager.spawnPepper(
        client_id,
        [0, 0, 0],
        [0, 0, 0, 1],
        spawn_ground_plane=True)

Or via the PepperVirtual class:

import time
import pybullet as p
import pybullet_data
from qibullet import PepperVirtual

if __name__ == "__main__":
    physicsClient = p.connect(p.GUI)
    p.configureDebugVisualizer(p.COV_ENABLE_RENDERING, 1)
    p.configureDebugVisualizer(p.COV_ENABLE_RGB_BUFFER_PREVIEW, 0)
    p.configureDebugVisualizer(p.COV_ENABLE_DEPTH_BUFFER_PREVIEW, 0)
    p.configureDebugVisualizer(p.COV_ENABLE_SEGMENTATION_MARK_PREVIEW, 0)

    p.setRealTimeSimulation(1)
    p.setGravity(0, 0, -9.81)

    p.setAdditionalSearchPath(pybullet_data.getDataPath())
    p.loadMJCF("mjcf/ground_plane.xml")

    pepper = PepperVirtual()
    pepper.loadRobot([0, 0, 0], [0, 0, 0, 1], physicsClientId=physicsClient)

More snippets can be found in the examples folder:

• A basic usage of the PepperVirtual class
• Launch multiple simulation instances
• A script using several simulation instances to compute the error on the joint positions
• A basic usage of the PepperRosWrapper class (bridge between qibullet and ROS)
• Script allowing a simulated model to mimic a real robot's movements

Documentation

The qibullet API documentation can be found here. The documentation can be generated via the following command (the epydoc Python package has to be installed beforehand, and the docs folder has to exist):

epydoc --html qibullet/ -o docs/api --name qibullet

Troubleshooting

OpenGL driver

If you encounter the message:

Workaround for some crash in the Intel OpenGL driver on Linux/Ubuntu

Your computer is using the Intel OpenGL driver. Go to Software & Updates, Additional Drivers, and select a driver corresponding to your GPU.