hanruihua/ir-sim

What it solves

IR-SIM is a lightweight, Python-based robot simulator designed to lower the barrier for prototyping robotics algorithms. It eliminates the need for complex coding or expensive hardware when testing navigation, control, and learning algorithms in custom 2D scenarios.

How it works

The simulator uses a declarative, YAML-driven approach to define the world, robot kinematics, sensors, and behaviors. Users configure their environment in a YAML file and run the simulation loop in Python. It includes a Matplotlib-based renderer for visualization and built-in collision detection.

Who it’s for

It is primarily aimed at academic researchers and students in robotics and AI, specifically those working on multi-agent systems, robot learning, and autonomous navigation.

Highlights

• YAML Configuration: Quickly set up scenarios without writing extensive code.
• Diverse Kinematics: Supports differential drive, omnidirectional, and Ackermann steering robots.
• Comprehensive Sensors: Includes 2D LiDAR, FMCW LiDAR, contact sensors, and FOV detectors.
• Advanced Collision Models: Features rigid-body contact (pushing boxes with mass and friction) and various collision avoidance behaviors like RVO, ORCA, and Social Force Model.
• Integration: Used in numerous academic publications for reinforcement learning and motion planning research.