Implementation:ARISE Initiative Robosuite Make

Metadata

Property	Value
Sources	robosuite
Domains	Robotics_Simulation, Software_Architecture
Last Updated	2026-02-15 12:00 GMT

Overview

Concrete factory function for instantiating robosuite simulation environments by name provided by the robosuite environments module.

Description

The robosuite.make() function creates simulation environments by looking up env_name in REGISTERED_ENVS and instantiating the matched class. Supports all MujocoEnv keyword arguments including robot selection, controller configuration, rendering options, and simulation parameters.

Usage

Import and call as the primary way to create any robosuite environment. Accepts environment name plus configuration kwargs.

Code Reference

Property	Value
Source	robosuite
File	robosuite/environments/base.py
Lines	L23-42 (re-exported at robosuite/__init__.py)
Import	import robosuite then robosuite.make(...) or from robosuite import make

Signature

def make(env_name, *args, **kwargs):
    """
    Instantiates a robosuite environment.

    Args:
        env_name (str): Name of the robosuite environment to initialize
        *args: Additional arguments
        **kwargs: Additional keyword arguments

    Returns:
        MujocoEnv: Desired robosuite environment

    Raises:
        Exception: [Invalid environment name]
    """

I/O Contract

Inputs

Parameter	Type	Required	Description
env_name	str	Yes	Registered environment name (e.g., "Lift", "Stack", "NutAssembly")
robots	str or list	No	Robot name(s) to use (e.g., "Panda", ["Panda", "Sawyer"])
controller_configs	dict	No	Controller configuration from load_composite_controller_config
has_renderer	bool	No	Enable on-screen rendering (default: False)
has_offscreen_renderer	bool	No	Enable off-screen rendering (default: True)
use_camera_obs	bool	No	Include camera images in observations (default: False)
control_freq	int	No	Controller frequency in Hz (default: 20)
horizon	int	No	Maximum episode steps (default: 1000)
hard_reset	bool	No	Full model reload on reset (default: True)
reward_shaping	bool	No	Use dense rewards instead of sparse (default: False)

Outputs

Return Value	Type	Description
Environment instance	MujocoEnv	Task-specific subclass instance (e.g., Lift, Stack, NutAssembly)

Usage Examples

Basic Example

Creating a basic Lift environment with a Panda robot:

import robosuite as suite

# Create environment with default settings
env = suite.make(
    env_name="Lift",
    robots="Panda",
    has_renderer=True,
    has_offscreen_renderer=False,
    use_camera_obs=False,
)

# Reset environment
obs = env.reset()

# Execute actions
for i in range(1000):
    action = env.action_spec[0].sample()
    obs, reward, done, info = env.step(action)
    env.render()

    if done:
        obs = env.reset()

env.close()

Advanced Example

Creating an environment with full configuration including custom controller:

import robosuite as suite
from robosuite.controllers import load_composite_controller_config

# Load controller configuration
controller_config = load_composite_controller_config(
    controller="OSC_POSE",
    robot="Panda"
)

# Create environment with advanced settings
env = suite.make(
    env_name="Stack",
    robots=["Panda", "Sawyer"],                    # Multi-robot setup
    controller_configs=controller_config,          # Custom controller
    has_renderer=True,                             # On-screen rendering
    has_offscreen_renderer=True,                   # Off-screen rendering
    use_camera_obs=True,                           # Include camera observations
    camera_names=["agentview", "robot0_eye_in_hand"],
    camera_heights=256,
    camera_widths=256,
    control_freq=20,                               # 20 Hz control
    horizon=1000,                                  # 1000 steps per episode
    reward_shaping=True,                           # Dense rewards
    hard_reset=False,                              # Faster resets
)

# Reset and get observation
obs = env.reset()

# Observation dictionary contains:
# - Robot joint positions and velocities
# - Object positions and orientations
# - Camera images (if use_camera_obs=True)
print("Observation keys:", obs.keys())

# Step through environment
action = env.action_space.sample()
obs, reward, done, info = env.step(action)

env.close()

Querying Available Environments

from robosuite import ALL_ENVIRONMENTS

# List all registered environments
print("Available environments:")
for env_name in ALL_ENVIRONMENTS:
    print(f"  - {env_name}")

# Output:
#   - Lift
#   - Stack
#   - NutAssembly
#   - PickPlace
#   - Door
#   - Wipe
#   - TwoArmLift
#   - ...

Related Pages

• Principle:ARISE_Initiative_Robosuite_Environment_Factory
• Environment:ARISE_Initiative_Robosuite_MuJoCo_Python