wg_robots_gazebo: gazebo_benchmarks | icra_navigation_gazebo | pr2_2dnav_gazebo | pr2_arm_gazebo | pr2_bringup_gazebo_demo | pr2_build_map_gazebo_demo | pr2_doors_gazebo_demo | pr2_gazebo_wg | pr2_kinect_test_description | pr2_laban_gazebo_demo | pr2_pickup_object_demo | pr2_plugs_gazebo_demo | pr2_simulator_benchmarks | pr2_sr_hand_gazebo_demo | pr2_sushi_sim | pr2_tabletop_manipulation_gazebo_demo | ros_epic_fail | texas_gazebo
- Code API
- diamondback - Msg API
- diamondback - Tutorials
- FAQ
- Reviews (na)
Used by (1)
Package Summary
Package for basic physics tests
- Author: John Hsu
- License: BSD
- Repository: wg-ros-pkg
- Source: svn https://code.ros.org/svn/wg-ros-pkg/stacks/wg_robots_gazebo/trunk/gazebo_benchmarks
Contents
Running CUDA Demos
CUDA Device Driver
You can do a system install of the CUDA device drivers for your graphics card. This step is currently manual, pending work from cuda package. Not to worry if the CUDA device drivers are not installed, parallel_quickstep by default compiles and runs off the CPU.
If your CUDA does not support doubles operations, you have to disable doubles in ODE before compiling.
C-Turtle Installation Instructions
Install ROS C-Turtle by following instructions here and be sure to install the PR2 All variant.
Make sure you have rosinstall installed and working. Save the rosinstall file in your home directory as simulator_overlay.rosinstall. Next, type
rosinstall ~/simulator_overlay /opt/ros/cturtle ~/simulator_overlay.rosinstall
to install in the ~/simulator_overlay directory.
Setup the ROS environment variables for your new overlay:
source ~/simulator_overlay/setup.sh
you might want to append the above line to your .bashrc for convenience.
Disabling Doubles
For CUDA devices that do not support doubles operations (compute capability version less than 1.3), compile ODE using floats,
roscd opende vi Makefile.opende.tarball
remove --enable-double-precision in CFG_OPTIONS.
Compile
To compile, type
rosmake gazebo_benchmarks
Running Benchmark Demos
Try:
roslaunch gazebo_benchmarks box_stack_block_cuda.launch






