Autonomous Drones in Industry: Drone-Controlled Scanning


More and more companies are relying on drones: According to a recent study by the German Aviation Association (BDL), the German drone market will increase from 840 million € to over 1.6 billion € by 2025 – this corresponds to an annual average growth rate of 14.5 percent. This development is mainly driven by the commercial market. Today, only one in nine drones is operated commercially in Germany, by 2025 it will be one in three drones.    Crucial factors are the rapid technical developments in the field of drone technology as well as the diverse applications for so-called Unmanned Aerial Vehicles (UAVs) or Remotely Piloted Aerial Systems (RPAS). Their advantages are obvious: Thanks to high-resolution and variable camera systems, drones deliver detailed images from different angles in a compact, mobile, and precise manner. Drones can also be extended by a wide variety of functions such as a GPS chip, special sensor, infrared, or measurement technology. The high-resolution images and data can serve as the basis for exact 3D models.

The drones with four or more rotors are usually piloted by remote control or from a ground control station. However, drones can only fully develop their potential if they can be used autonomously. But what exactly is meant when we talk about autonomy? How autonomous are drones currently?

The six Levels of Autonomy

The autonomous capabilities of drones can be divided into different levels – comparable to autonomous driving. The model of the European Cockpit Association (ECA) is based on the rules from the automotive industry according to SAE J3016, which have been transferred to UAVs.



The definition of the term autonomy is fundamental to this model: A clear distinction is made between an autonomous aircraft and an autonomous operation. An autonomous aircraft does not allow pilot intervention in the management of the flight. It can completely determine its actions (including destination and route) and is free of external control or influence – keyword artificial intelligence (AI). In an autonomous operation, a drone is remotely piloted by a human – but is operating without pilot intervention in the management of the flight. This is where the term automation comes into play: The autonomy of a drone always depends on the automation technology and the (un)necessity for manual control.

A high degree of automation is therefore often equated with the term "autonomous operation" because human intervention is reduced to a minimum.

The ECA model distinguishes six different levels of automation from 0 to 5. The number reflects the degree of automation: from level 0 – no automation at all – to level 5 – complete automation. While the drone is controlled by the Pilot-in-Command during the entire operation at level 0, in level 1 the drone contributes to controlling one of the three axes of flight and detecting obstacles.

At level 2 drones can fly independently on all three axes and warn of obstacles. At level 3, the pilot takes a back seat, and the drone performs tasks autonomously (for example, measurements). Obstacles are detected under defined conditions and the drone avoids them autonomously. The pilot still has to react to system errors and take control as a backup if necessary. Level 4 already represents a high level of automation. The drone can perform predetermined tasks completely on its own, while the human plans and monitors the flight. The drone reacts to errors by resorting to backup systems or solving problems autonomously. Complete automation at the fifth level is equivalent to an autonomous or "smart" drone – the drone can autonomously plan and execute all parameters of a flight mission with the help of artificial intelligence (AI). Man merely gives the flight command.

Applications for autonomously flying Drones

The applications for drones in industry are diverse: Drones are particularly suitable for hard-to-reach buildings and extensive terrain. They are used for the observation, monitoring and verification of plants and their construction progress. Subjects can be, for example, catenary, pipelines, solar systems, or wind turbines. Precise surveying data is required for the planning of construction areas, the route mapping of traffic routes and supply lines and in urban building construction and civil engineering. Drones with high-resolution cameras and sensors are already being used to create inch-perfect terrain maps.

To document the work progress, the drone can always head to the exactly the same points at the same altitude at different times. The drones can already fly predefined, programmed routes autonomously, including automated take-off and landing. Humans can observe the mission live on the monitor. Some drones can detect spontaneous obstacles and avoid them.

In addition, there are drones that can operate without satellite-based navigation. This is one of the conditions for using drones in enclosed spaces such as factory halls – where the GPS signal fails. Here, drones can be used to measure the production halls and thus significantly simplify and accelerate the factory planning process. Other possible tasks include inspection, inventory, and monitoring.

If one compares the examples mentioned with the classification of the ECA, it becomes clear that when we talk about autonomous drones today, we usually mean high or fully automated drone flights. In other words, the drone can navigate, fly or hover autonomously without a pilot and completing tasks – for example, by scanning its surroundings. This capability is equivalent to level 3 on the ECA automation scale.

The advantages of drones for industry are obvious: they deliver data faster than conventional methods, even in hard-to-reach areas. Drone-based image evaluation and mapping enables a high data density. For example, high-resolution point clouds can be created with the help of 3D laser scanning. But how exactly does 3D mapping with drones work?   In our next blog post, we will introduce you to the two most important technologies used in mapping with (autonomous) drones: photogrammetry and LiDAR.