Heterogeneous robots equipped with multi-modal sensors (e.g., UAV, wheeled and legged terrestrial robots) provide rich and complementary functions that may help human operators to accomplish complex tasks in unknown environments. However, seamlessly integrating heterogeneous agents and making them interact and collaborate still arise challenging issues. In this paper, we define a ROS 2 based software architecture that allows to build incarnated heterogeneous multi-agent systems (HMAS) in a generic way. We showcase its effectiveness through a scenario integrating aerial drones, quadruped robots, and human operators (see https://youtu.be/iOtCCticGuk). In addition, agent spatial awareness in unknown outdoor environments is a critical step for realizing autonomous individual movements, interactions, and collaborations. Through intensive experimental measurements, RTK-GPS is shown to be a suitable solution for achieving the required locating accuracy.

翻译：配备多模态传感器（如无人机、轮式和腿式地面机器人）的异构机器人提供了丰富的互补功能，可协助人类操作员在未知环境中完成复杂任务。然而，无缝集成异构智能体并使其交互与协作仍面临诸多挑战。本文定义了一种基于ROS 2的软件架构，能够以通用方式构建具身化异构多智能体系统（HMAS）。我们通过一个集成空中无人机、四足机器人和人类操作员的场景（参见https://youtu.be/iOtCCticGuk）展示了其有效性。此外，在未知室外环境中实现智能体空间感知是自主个体运动、交互与协作的关键步骤。通过大量实验测量，RTK-GPS被证明是实现所需定位精度的合适方案。