HERCULES: An Open-Source Simulation Framework for Heterogeneous Multi-Robot SLAM, Collaborative Perception, and Exploration

Modern robotics missions, such as disaster response and environmental monitoring, increasingly rely on heterogeneous teams of Unmanned Aerial Vehicles (UAVs) and Unmanned Ground Vehicles (UGVs). While UAVs provide rapid aerial coverage, UGVs offer persistent ground-level sensing and higher payload capacity. Despite this synergy, existing simulation platforms often struggle to support concurrent aerial-ground operations at scale. Most simulators are either optimized for single-vehicle types, lack the photorealistic environments required for perception research, or fail to provide the synchronized data-collection tools necessary for developing robust multi-robot coordination algorithms.

HERCULES (HEterogeneous multi-Robot simulator for Coordination, scene Understanding, Large-scale Exploration and SLAM) is built upon Unreal Engine 5 (UE5) and extends the capabilities of AirSim and Cosys-AirSim. The authors re-architected the simulation core to resolve a fundamental conflict where previous frameworks restricted sessions to a single vehicle type. This allows UAVs and UGVs to operate concurrently within a shared world state and a unified simulation clock.

Key features include:

• Unified Navigation Stack: A shared waypoint-tracking interface and kinodynamic planner that allows high-level commands to be issued to both aerial and ground platforms.
• Advanced Sensing: Integration of physics-based long-wave infrared (LWIR) cameras and night-vision modes, enabling research in degraded visual environments.
• Dynamic Environments: Support for intelligent agents (pedestrians, traffic, wildlife) and environmental phenomena (fire, flooding, disease spread) that can be configured via robotics-ready APIs.
• Data Collection: A robust pipeline that supports both passive dataset generation (replaying trajectories) and active closed-loop simulation (online planning).

HERCULES bridges the gap between high-fidelity game-engine rendering and the rigorous requirements of robotics research. By providing an open-source, out-of-the-box testbed for heterogeneous multi-robot SLAM and collaborative perception, it enables researchers to prototype and validate complex coordination strategies before deploying them in the real world. The release includes a benchmark dataset spanning desert, forest, and city environments, providing a standardized way to evaluate multi-robot autonomy.