MIT mini cheetah quadruped robot simulated in pybullet environment using ros.
MIT mini cheetah quadruped robot simulated in pybullet environment using ros.
MIT mini cheetah use customized simulator and lcm framework, which is not a popular way to do the robot development. Now, we extract the algorithm and do the simulation using ros and pybullet. This can be simple to deploy the system into different custom robot or plantform, and easy to learn the algorithm.
Ubuntu 18.04, ROS Melodic
use Logitech gamepad to control robot
git clone https://github.com/Derek-TH-Wang/gamepad_ctrl.gitmsg rospack and rviz plugin
https://github.com/loco-3d/whole_body_state_msgs.git
https://github.com/eborghi10/whole_body_state_rviz_plugin.gitcd {your workspace}
catkin make
source devel/setup.bashpip3 install -r requirements.txtyou can modify the config/quadruped_ctrl_cinfig.yaml/terrain to deploy different terrains, there are four terrains supported in the simulator now, for example:
"plane"
"stairs"
"random1"
"random2"
"racetrack"run the gamepad node to control robot:
roslaunch gamepad_ctrl gamepad_ctrl.launchrun the controller in simulator:
roslaunch quadruped_ctrl quadruped_ctrl.launchswitch the camera on / off:
camera set True or False in config/quadruped_ctrl_config.yaml, then launch the rviz to see the point cloud:
roslaunch quadruped_ctrl vision.launchalso can switch the gait type:
rosservice call /gait_type "cmd: 1"gait type:
0:trot
1:bunding
2:pronking
3:random
4:standing
5:trotRunning
6:random2
7:galloping
8:pacing
9:trot (same as 0)
10:walking
11:walking2No open issues yet, or sync has not completed.