agent-sim-bridge

Simulation-to-reality bridge for AI agents with environment adapters

Part of the AumOS open-source agent infrastructure portfolio.

Features

• TransferBridge applies a CalibrationProfile (per-dimension linear scale + offset + noise model) to translate observations and actions between simulator coordinates and real-world distributions
• Calibrator fits calibration profiles by regression on paired sim/real observation datasets, with support for iterative recalibration as new data arrives
• DomainRandomization perturbs simulation parameters (friction, mass, sensor noise, lighting) across configurable ranges to produce policies that transfer without fine-tuning
• Simulation scenario recorder and replay engine lets you capture real-world trajectories, replay them in simulation, and measure the fidelity gap
• SafetyMonitor enforces physical and policy constraints during both simulated and real execution, with configurable boundary definitions and emergency stop hooks
• Sensor fusion module combines readings from multiple simulated or real sensors with configurable noise injection and covariance-weighted fusion
• Backend adapters for Gazebo and PyBullet expose a uniform SimulationEnvironment interface so agents can be tested against either simulator without code changes

Quick Start

Install from PyPI:

pip install agent-sim-bridge

Verify the installation:

agent-sim-bridge version

Basic usage:

import agent_sim_bridge

# See examples/01_quickstart.py for a working example

Documentation

• Architecture
• Contributing
• Changelog
• Examples

Enterprise Upgrade

For production deployments requiring SLA-backed support and advanced integrations, contact the maintainers or see the commercial extensions documentation.

Contributing

Contributions are welcome. Please read CONTRIBUTING.md before opening a pull request.

License

Apache 2.0 — see LICENSE for full terms.

Part of AumOS — open-source agent infrastructure.