Understanding drone sensor data for autonomous flight

Flare Bright是一家世界领先的航空航天无人机和先进空中机动性（AAM）软件公司，使用先进的数字双胞胎（DT）开发机器学习（ML）增强的软件生态系统，该系统将实现安全、可靠、自主的超视距（BVLOS）飞行。Flare Bright的核心ML-DT软件生态系统正在开发两种重要的商业产品，Flare Bright拥有巨大的潜在市场，并被认为是无人机技术实现经济增长的关键推动因素。首先，Flare Bright开发了一种高分辨率的“无传感器”风力测量能力，该能力使用无人机本身作为传感器。其他解决方案，如皮托管探头，不太准确，响应速度较慢，从尺寸和重量的角度来看不实用。这种独特的能力已经在卡迪夫机场的UKRI未来飞行项目SafeZone中得到了证明，并且在ICAS 2022技术论文中对所达到的精度进行了详细审查。这种测量能力的准确性和分辨率使其成为更安全飞行的关键推动因素，允许更灵敏的飞行控制调谐并生成风数据，从而能够在飞机所经历的风条件与当前可用的平均预测数据显著不同的城市或复杂流动环境中定义安全操作限制。其次，Flare Bright开发了一种ML助推惯性导航系统（INS），当GPS被拒绝或损坏时，使用现有的小型廉价机载传感器提供准确的短期导航。现有的高精度惯性导航系统体积大、重量重、功耗大，不适合未来的无人机和电动飞机，因为重量和功率是关键。因此，我们的能力是BVLOS的关键，因为GPS丢失已经经常阻碍操作，并且在未来，由于技术使用的增加和故意干扰破坏信号的电磁干扰，可能会更频繁地发生。该项目旨在解决支撑这两种产品的挑战。Flare Bright开发了一种正在申请专利的技术，该技术可以使用嵌入式DT来估计飞机上的气流，如果地面速度已知，则可以从中估计风速，反之亦然。然而，当应用于广泛的飞行包线，这种技术是由一个复杂的“一对多”的问题与多个潜在的解决方案阻碍。该项目使用最先进的数学技术，旨在确定是否可以确定或计算出单个解决方案比其他解决方案更有可能，从而显着提高Flare Bright核心产品的影响力和可行性。

Flare Bright is a world-leading aerospace drone and Advanced Air Mobility (AAM) software company developing a Machine Learning (ML) boosted software ecosystem using advanced Digital Twins (DT) that will enable safe, reliable, autonomous Beyond Visual Line Of Sight (BVLOS) flight.Flare Bright's core ML-DT software ecosystem is leading to two commercially important products, both with huge potential markets and identified as critical enablers to drone technology achieving the economic growth predicted.Firstly, Flare Bright has developed a high resolution 'sensorless' wind measurement capability that uses the drone itself as the sensor. Other solutions, such as pitot probes, are less accurate, slower in response and not practical from a size and weight perspective. This unique capability has been demonstrated at Cardiff Airport, within UKRI's Future Flight project SafeZone, and the accuracy achieved has been scrutinised in an ICAS 2022 technical paper. The accuracy and resolution of this measurement capability makes it a key enabler in safer flight, allowing more responsive flight control tunes and generating wind data that would enable safe operating limits to be defined in urban or complex flow environments where wind conditions experienced by aircraft are significantly different from averaged forecast data that is currently available.Secondly, Flare Bright has developed an ML-boosted Inertial Navigation System (INS) that provides accurate short-term navigation when GPS is denied or corrupted using existing, small and inexpensive, on-aircraft sensors. Existing high accuracy INS are large, heavy and consume lots of power, making them unsuitable for drones and electric aircraft of the future, where weight and power are critical. Our capability is therefore key to BVLOS as GPS drop-outs already regularly hinder operation, and, in future, are likely to occur more frequently due to electromagnetic interference from increased technology use and deliberate jamming disrupting signals.This project aims to address a challenge underpinning both these products. Flare Bright has developed a patent pending technique that enables airflow over the aircraft to be estimated using an embedded DT, from which either wind speed can be estimated if ground velocity is known or vice versa. However, when applied across a wide flight envelope, this technique is hindered by a complex "one-to-many" problem with multiple potential solutions. Using cutting edge mathematical techniques, this project aims to identify if a single solution may be ascertained or calculated to be more likely than the others, thereby significantly increasing the impact and viability of both Flare Bright's core products.