IR-SIM is an open-source, Python-based, lightweight robot simulator designed for navigation, control, and learning. It provides a simple, user-friendly framework with built-in collision detection for modeling robots, sensors, and environments. Ideal for academic and educational use, IR-SIM enables rapid prototyping of robotics and learning algorithms in custom scenarios with minimal coding and hardware requirements.

• Simulate robot platforms with diverse kinematics, sensors, and behaviors (support).
• Quickly configure and customize scenarios using straightforward YAML files. No complex coding required.
• Visualize simulation outcomes using a naive visualizer matplotlib for immediate debugging.
• Support collision detection and customizable behavior policies for each object.
• Suitable for mutli-agent/robot learning (Projects).

Multi-Robot RVO Collision Avoidance Source	Ackermann Robot with 2D LiDAR Source	HM3D / MatterPort3D Grid Map Source
Field-of-View Detection Source	Dynamic Random Obstacles Source	200-Agent ORCA via pyrvo Source

Requires Python >= 3.10

pip install ir-sim

# Optional: keyboard control and all extras
pip install ir-sim[all]

git clone https://github.com/hanruihua/ir-sim.git
cd ir-sim
pip install -e .

git clone https://github.com/hanruihua/ir-sim.git
cd ir-sim
uv sync

A minimal example: a differential-drive robot navigates toward a goal using the built-in dash behavior.

import irsim

env = irsim.make('robot_world.yaml') # initialize the environment with the configuration file

for i in range(300): # run the simulation for 300 steps

    env.step()  # update the environment
    env.render() # render the environment

    if env.done(): break # check if the simulation is done

env.end() # close the environment

YAML Configuration: robot_world.yaml

world:
  height: 10  # the height of the world
  width: 10   # the width of the world
  step_time: 0.1  # 10Hz calculate each step
  sample_time: 0.1  # 10 Hz for render and data extraction
  offset: [0, 0] # the offset of the world on x and y

robot:
  kinematics: {name: 'diff'}  # omni, omni_angular, diff, acker
  shape: {name: 'circle', radius: 0.2}  # radius
  state: [1, 1, 0]  # x, y, theta
  goal: [9, 9, 0]  # x, y, theta
  behavior: {name: 'dash'} # move toward to the goal directly
  color: 'g' # green

For more examples, see the usage directory and the documentation.

• English: https://ir-sim.readthedocs.io/en
• Chinese (中文): https://ir-sim.readthedocs.io/zh-cn

• [RAL & ICRA 2023] rl-rvo-nav -- Reinforcement learning-based RVO behavior for multi-robot navigation.
• [RAL & IROS 2023] RDA_planner -- Accelerated collision-free motion planner for cluttered environments.
• [T-RO 2025] NeuPAN -- Direct point robot navigation with end-to-end model-based learning.

• DRL-robot-navigation-IR-SIM -- Deep reinforcement learning for robot navigation.
• AutoNavRL -- Autonomous navigation using reinforcement learning.
• IRSIM-3DGS-Bridge -- A closed-loop bridge from 3D Gaussian Splatting scenes to IR-SIM planning/following and back to Habitat-GS trajectory playback.

If you find IR-SIM useful, please consider starring ⭐ this project and citing our paper:

@misc{han2026irsimlightweightskillnativesimulator,
      title={IR-SIM: A Lightweight Skill-Native Simulator for Navigation, Learning, and Benchmarking},
      author={Ruihua Han and Shuai Wang and Chengyang Li and Rui Gao and Xinyi Wang and Zhe Liu and Guoliang Li and Yupu Lu and Qi Hao and Jia Pan and Hengshuang Zhao},
      year={2026},
      eprint={2606.08729},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2606.08729},
}

Contributions are welcome! Please see CONTRIBUTING.md for guidelines.

• PythonRobotics

IR-SIM is released under the MIT License.