A collection of robot models tailored to robosuite.
This repository also provide examples of how to create a custom composable robot model with robosuite.
See compositional.py: how to compose a new robot model.
Robot Model | Names |
---|---|
RobotBaseModel |
aloha_mount, b1_base, go2_base |
GripperModel |
ability_hands, aloha_gripper, g1_three_finger_gripper, pr2_gripper, schunk_svh_hands, umi_gripper, yumi_gripper, z1_gripper |
ManipulatorModel |
aloha_robot, arx5_robot, g1_robot, h1_robot, pr2_robot, vx330s_robot, yumi_robot, z1_robot |
Example code:
from robosuite.robots import register_robot_class
from robosuite.models.robots import Panda
import robosuite as suite
from robosuite.controllers import load_composite_controller_config
import mujoco
@register_robot_class("WheeledRobot")
class MobilePanda(Panda):
@property
def default_base(self):
return "OmronMobileBase"
@property
def default_arms(self):
return {"right": "Panda"}
# Create environment
env = suite.make(
env_name="Lift",
robots="MobilePanda",
controller_configs=load_composite_controller_config(controller="BASIC"),
has_renderer=True,
has_offscreen_renderer=False,
render_camera="agentview",
use_camera_obs=False,
control_freq=20,
)
# Run the simulation, and visualize it
env.reset()
mujoco.viewer.launch(env.sim.model._model, env.sim.data._data)
To load a robot model to see what it looks like, you can use the following command:
python examples/load_compositional_robot_example.py --robots UR5eOmron --controller BASIC
Run pre-commit run --all-files
before open a pull request.