Distributed Autonomous Pilot Control for Unmanned Aerial Vehicles (UAVs)

无人驾驶飞行器（UAV）等自动驾驶车辆在各种应用中得到越来越多的认可。无人机行业将在数量和应用领域方面出现爆炸式增长，而自动或半自动驾驶汽车将在可预见的未来进入市场。可以观察到，随着无人机的自主或半自主操作变得越来越流行和必要，自动驾驶仪控制的趋势正在走向智能化。因此，如何开发基于分布式传感器或传感器网络的智能自动驾驶控制器是无人机行业的重要兴趣和技术创新。在这方面，一个现场传感器网络，如摄像头、声波传感器、雷达、激光雷达等，必须将信息融合在一起，以提供对周围环境的感知。然后，自动驾驶仪控制器可以根据融合的信息做出决策来操纵无人机的自主操作。在这个项目中，打算1。开发分布式三维重建算法，用于相机和激光雷达的传感和感知；2. 开发网络环境下与飞行自动驾驶仪控制集成的传感数据融合机制；和3。在成本和效率评估的基础上，采用先进的硬件技术，如带图形处理单元（GPU）的多处理器或FPGA解决方案来实现控制器。特别是，将为现场传感器节点开发分布式算法，并为网络化和互补配置的现场传感器探索和开发数据融合机制，以便以自动化的方式组装每个单独传感器收集的信息。基于这种分布式结构和数据融合机制，将构建无人机自主飞行的智能自动驾驶控制器。预计该项目成果将应用于优化各种自主任务的智能自动驾驶仪设计。

Autonomous vehicles such as Unmanned Aerial Vehicle (UAV) are gaining increasing acceptance in variousapplications. The UAV industry will see explosive growth in terms of both quantities and the areas ofapplications while the autonomous or semi-autonomous vehicles will be in the market in foreseeable future. Itis observed that the trend of autopilot control is going towards intelligent as autonomous or semi-autonomousoperations of UAVs become more and more popular and necessary. Therefore, how to develop distributedsensors or sensor network based intelligent autopilot controllers is of vital interest and also technology-wiseinnovative for the UAV industry. In this regard, a network of field sensors such as cameras, sonic sensor,RADAR, LiDAR, etc., has to be applied with information fused together to provide the perception of thesurrounding environment. The autopilot controller could then make decision based on the fused information tomaneuver the autonomous operation of UAVs.In this project, it is intended to 1. develop distributed 3-D reconstruction algorithms for sensing and perceptionperformed by cameras and LiDARs; 2. develop sensing data fusion mechanism in a networked environment tobe integrated with the flight autopilot control; and 3. implement the controller with advanced hardwaretechnology such as multi-processors with graphic processing unit (GPU) or FPGA solution upon cost andefficiency assessment. In particular, the distributed algorithms will be developed for field sensor nodes and thedata fusion mechanism will be explored and developed for a networked and complimentary configuration offield sensors in order to assemble the information collected by each individual sensor in an automated way.Based on this distributed structure and the data fusion mechanism, the intelligent autopilot controller will bebuilt up for autonomous operation of UAVs. It is expected that the project deliverable be applied to optimizeintelligent autopilot design for various autonomous missions.