FollowPV - Developing autonomous unmanned aerial vehicles with spatial awareness for improved image quality from solar farm inspections

FollowPV项目计划开发一种“自动驾驶”（半自动）无人机系统，用于检查太阳能发电场。我们的设备将允许无人机跟随一排排太阳能电池板，就像“自动驾驶”汽车能够保持在车道上一样。然而，与汽车不同的是，无人机不是通过车轮与道路相连的。因此，我们的设备还必须使无人机能够在不平坦的地形上跟随太阳能电池板的升降。太阳能发电场对英国的能源供应和减少排放至关重要，因此需要定期检查有缺陷的部件。我们使用带有专业摄像头的无人机一次检查整个太阳能农场，这比步行检查面板更有效。这降低了太阳能农场的维护成本，使其能够在最佳状态下运行，这有助于降低消费者的电力成本。然而，有些缺陷只能在近距离观察到，却会揭示出未来早期的系统性退化问题。目前的无人机不够精确，无法非常接近太阳能电池板，因此有时仍需要人工检查。这些方法非常耗时、昂贵，还涉及健康和安全风险。为了使用无人机捕捉这种超高细节图像，我们想飞得更接近面板（在5米内）。然而，就像“卫星导航”不够精确，无法控制自动驾驶汽车的方向盘一样，GPS也不够精确，无法控制离太阳能电池板如此之近的无人机。为了准确地做到这一点，无人机（就像汽车一样）需要能够“看到”它的环境，并理解和使用这些信息来进行微小的控制调整。这需要无人机上的特殊传感器和机载人工智能（AI），后者可以快速处理并进行飞行纠正。拉夫堡大学（LU）和埃塞克斯大学（UoE）已经拥有将无人机技术用于“智能农业”（例如作物病害监测）的专业知识，但类似的技术可以应用于太阳能农场。在我们拟议的合作伙伴关系中，陆航和UoE在无人机自动化方面的专业知识将与_Above_在太阳能发电场检查方面的专业知识以及国际客户和商业合作伙伴的全球网络相结合。最终，我们对这个项目的愿望是确保英国和世界上的太阳能发电厂尽可能高效地工作，从而减少我们对化石燃料的依赖。

The FollowPV project plans to develop a 'self-driving' (semi-automated) drone system for inspecting solar farms. Our device will allow a drone to follow rows of solar panels in the same way that a 'self-driving' car is able to keep in lane. However, unlike a car, a drone is not connected to the road by wheels. Therefore, our device must also enable the drone to follow the rise and fall of solar panels over uneven terrain.Solar farms are critical to the UK's energy supply and to reducing emissions, so they need to be inspected regularly for defective components. We use drones with specialist cameras to inspect entire solar farms in a single visit, which is more efficient than inspecting panels on-foot. This reduces maintenance costs of solar farms allowing operation at optimum condition, which helps keep down the cost of electricity to the consumer.However, some defects are only visible very close-up, yet reveal early systemic degenerative problems for the future. Current drones are not accurate enough to fly very close to solar panels, and therefore manual inspections are sometimes still needed. These are very time-consuming, expensive, and involve health and safety risk.To use a drone to capture this ultra-high detail imagery, we want to fly much closer to the panels (within 5m). However, in the same way that 'sat nav' is not accurate enough to control the steering wheel of a self-driving car, then GPS is not accurate enough to control a drone so near to the solar panels. To do this accurately, the drone (like the car) needs to be able to 'see' its environment, and to understand and use this information to make tiny control adjustments. This requires special sensors on the drone, and onboard artificial intelligence (AI), which can rapidly process and make in-flight corrections.Loughborough University (LU) and the University of Essex (UoE) already have expertise in utilising drone technology with this capability for use in 'smart agriculture' (e.g. crop disease monitoring), but similar technology can be applied to solar farms.In our proposed partnership, the expertise of LU and UoE in drone automation will be combined with _Above_'s expertise in solar farm inspection and worldwide network of international customers and commercial partners. Ultimately, our desire with this project is to ensure that the UK and the world's solar plants are working as efficiently as possible, thus reducing our reliance on fossil fuels.