In this study, we present a novel paradigm for industrial robotic embodied
agents, encapsulating an 'agent as cerebrum, controller as cerebellum'
architecture. Our approach harnesses the power of Large Multimodal Models
(LMMs) within an agent framework known as AeroAgent, tailored for drone
technology in industrial settings. To facilitate seamless integration with
robotic systems, we introduce ROSchain, a bespoke linkage framework connecting
LMM-based agents to the Robot Operating System (ROS). We report findings from
extensive empirical research, including simulated experiments on the Airgen and
real-world case study, particularly in individual search and rescue operations.
The results demonstrate AeroAgent's superior performance in comparison to
existing Deep Reinforcement Learning (DRL)-based agents, highlighting the
advantages of the embodied LMM in complex, real-world scenarios.

本研究提出了一个新的产业机器人具身化代理范例，其采用 “代理为大脑，控制器为小脑” 的架构。我们的方法利用大型多模态模型（LMMs）在名为 AeroAgent 的代理框架中，针对产业环境中的无人机技术进行了针对性的设计。为了实现与机器人系统的无缝集成，我们引入了 ROSchain，这是一个连接基于 LMM 的代理与机器人操作系统（ROS）的专用链接框架。通过广泛的实证研究，包括对 Airgen 的模拟实验和真实案例研究，特别是在个体搜救行动中，我们报告了研究结果。结果表明，与现有的基于深度强化学习（DRL）的代理相比，AeroAgent 在复杂的真实场景中表现出更高的性能，凸显了具身化 LMM 在复杂真实世界情境中的优势。

代理作为大脑，控制器作为小脑：在无人机上实现基于内容感知的代理

Agent as Cerebrum, Controller as Cerebellum: Implementing an Embodied  LMM-based Agent on Drones

The purpose of this paper is to develop an unmanned aerial vehicle (UAV)
using a quadcopter with the capability of video surveillance, map coordinates,
a deployable parachute with a medicine kit or a food pack as a payload, a
collision warning system, remotely controlled, integrated with an android
application to assist in search and rescue operations.
Applied research for the development of the functional prototype,
quantitative and descriptive statistics to summarize data by describing the
relationship between variables in a sample or population. The quadcopter
underwent an evaluation using a survey instrument to test its acceptability
using predefined variables to select respondents within Caloocan City and
Quezon City, Philippines.
Demographic profiles and known issues and concerns were answered by 30
respondents. The results were summarized and distributed in Tables 1 and 2.
In terms of demographic profiles, the number of SAR operators within the
specified areas is distributed equally, most are male, single, and within the
age bracket of 31 and above. In issues and concerns, the most common type of
search and rescue was ground search and rescue. Human error is the primary
cause of most injuries in operating units. The prototype was useful and
everyone agreed, in terms of acceptability, drone technology will improve
search and rescue operations.
The innovative way of utilizing Android and drone technology is a new step
towards the improvement of SAR operations in the Philippines.
The LiPo battery must be replaced with a higher capacity and the drone
operator should undergo a training course and secure a permit from the Civil
Aviation Authority of the Philippines (CAAP).

本研究旨在开发一种使用四旋翼无人机进行监视和救援的功能样机，该样机具备视频监控能力、地图坐标定位功能、可投放药箱或食品包的降落伞、碰撞预警系统，并与安卓应用程序集成以辅助搜救行动。研究利用应用型研究方法开发功能样机，并使用定量和描述性统计方法，通过描述样本或总体中变量之间的关系来总结数据。针对该样机进行了一项评估，使用调查工具测试了其可接受性，并使用现有变量在菲律宾卡卢坎市和奎松市选择了 30 位受访者进行回答和总结。结果表明在人口统计资料中，指定地区内搜救人员数量分布均匀，主要为男性、单身，年龄在 31 岁及以上。在问题和关注方面，最常见的搜救类型是地面搜救，操作错误是最多导致伤害的原因。该样机被认为是有用的，大家都同意，无人机技术将提高搜救行动的效果。利用安卓和无人机技术的创新方式是菲律宾搜救行动改进的新举措。需要用更高容量的锂聚合物电池替换，无人机操作员还应接受培训并获得菲律宾民航局的许可。