Model users of DJI aircraft are very afraid of encountering the "nightmare" of "compass interference". Whenever the APP prompts "compass interference", their mood immediately becomes very nervous, as if unpredictable danger is about to occur. In order to avoid "compass interference" problems, model users habitually calibrate their compass before flying, but even so, this prompt still appears in the air, which inevitably makes people angry and confused. Here, based on personal opinions, we will analyze the reasons for "compass interference" and its impact on the quality and safety of drone flight without a systematic approach.

　　(1) The core of upper level applications

　　To clarify the issue of compass interference, we should first talk about the GPS sensor.

　　In my opinion, GPS is simply the heart of various apps and smart hardware nowadays. With GPS, a series of location-based services have emerged: Baidu Maps uses GPS to locate and plan routes and recommend various entertainment services around them, WeChat/QQ recommends friends around them through location information, Uber/Didi recommends drivers and passengers around them through location information, Meituan/Nuomi recommends food around them through location information, and Ctrip/Tongcheng recommends hotels and attractions around them through location information, Various sports apps and intelligent hardware for sports use GPS to calculate the mileage, speed, and calories burned during exercise. Intelligent alarm hardware also monitors the loss of bicycles or children through GPS, and so on. So, GPS has given birth to a massive number of services, created countless job opportunities, and brought incredible changes to our lives.

　　(1.1) GPS based upper layer application of unmanned aerial vehicles

　　To get to the point, drones can be said to be a combination of apps and intelligent hardware, and naturally have a great relationship with GPS. Now let's take a look at what high-quality services GPS has brought to drone users in its upper level applications:

　　① Smooth hovering. Those who fly over the attitude mode know that the airplane will drift on the horizontal plane in this mode. Every time you take a photo, you will be distracted by controlling the airplane's movement and unable to focus on the beautiful scenery of the camera.

　　② Accurate return flight. Without GPS, mods will not be able to activate one click return when the plane is out of sight, allowing the plane to fly back smoothly and automatically, nor can it fly back when low voltage occurs again.

　　③ Low battery warning. Without GPS, the aircraft will not know its distance from the takeoff point, and cannot calculate the current remaining battery level in real time based on the distance, so it should return.

　　④ Ground station services. Without GPS, mods will not be able to guide the aircraft's waypoint on the ground station, and will not be able to automatically complete functions such as waypoint tracking and hotspot surround.

　　⑤ Follow Me. Without GPS, the Follow Me function based on GPS cannot be implemented.

　　⑥ APP trajectory display. Without GPS, model users will no longer be able to fly beyond line of sight by watching the trajectory of the aircraft on the app.

　　⑦ Various industry applications. Without GPS, various industries such as transportation, monitoring, and agriculture would also be unable to operate.

　　So, without GPS, drones would appear completely unintelligent.

　　(1.2) GPS based navigation system

　　Let's take a look at what information GPS, as a sensor, brings to the navigation system. A navigation system can be simply understood as a system that processes various sensor information to output information such as position, velocity, attitude, and heading. Nowadays, the navigation systems of drones are all integrated navigation systems, with integrated sensors including important inertial measurement units (IMUs) such as accelerometers and gyroscopes, compasses, GPS, visual sensors, radar, ultrasound, and so on. After adding GPS, the navigation system will obtain the following observation information: (1) position in the world coordinate system and (2) velocity in the world coordinate system.

　　With GPS position and velocity information, navigation systems can theoretically process this rough information, allowing drones to know their position and velocity in the world coordinate system.

　　But all of this is not that simple, please refer to the next section "(2) Drones have no sense of direction".

　　(2) Drones have no sense of direction

　　Give a common scene of asking for directions:

　　Ctrl: 'May I know how to get to Wanda Plaza?'

　　Navi: 'Walk 300 meters north and then 100 meters east to get there.'

　　Ctrl: 'Sorry, I don't have a sense of direction, I don't know where the north is.'

　　Navi: 'Okay, to put it another way, you go ahead 300m, then go right 100m and you'll be there.'

　　Ctrl: 'Understood, thank you.'

　　In this scene, when Navi guides Ctrl in an east-west, north-south manner, Ctrl does not understand because it lacks a sense of direction. But when Navi guides Ctrl in a forward, backward, left-right, left-right manner, Ctrl understands how to go. From this conversation, we can find the concept of drone coordinate system in the corresponding area. The coordinate system of drones generally has two coordinate systems: the "world coordinate system" and the "body coordinate system":

　　① The world coordinate system can be simply understood as a coordinate system defined in an east-west, north-south manner.

　　② The ontology coordinate system can be simply understood as a coordinate system located in a forward, backward, left-right manner.

　　GPS is in the world coordinate system, telling drones their position and speed in the world coordinate system. Unfortunately, drones also lack a sense of direction!

　　The favorite way for drones to move is to fly left or right, rather than east or north. So, in flight control, coordinate conversion must be performed to convert the "world coordinate system" to the "body coordinate system".

　　Returning to the scene of asking for directions, if Ctrl has a good sense of direction, he can immediately search for where the north direction is, and then convert Navi's north direction to forward direction based on Navi's first sentence. The same goes for drones, which need to find their relative direction in front and north in order to understand the position and speed in the world coordinate system. An important sensor that shoulders this responsibility is the compass. Speaking of this, it seems that we have just entered the main topic.

　　The compass of a drone may seem insignificant, but it actually plays a significant role. In order to enable various core services based on GPS, drones need to use a compass to connect "world coordinates" and "body coordinates".

　　(3) A single wooden bridge connecting two worlds

　　Heading is a single wooden bridge that connects the world coordinate system and the body coordinate system. If it deviates, it will fall into a vortex.

　　Let's give a scenario first:

　　A accidentally crashed in the desert and was able to survive. He decided to carry food on his back, choose a direction, and keep walking forward without turning, hoping to get out of the desert. However, a few days later he discovered the wreckage of an airplane. He despairingly found himself returning to the place of departure.

　　In this scene, A kept walking forward and finally made a detour back to the original position, for a simple reason. Although A keeps walking forward, if there is no fixed reference object in front to correct their step when walking, they will slowly deviate from the initially set front due to the slight difference in step length between their legs, and eventually walk out in an arc. People who walk with their eyes closed can also have a similar experience that if they don't take ten steps, the direction is actually crooked. In the desert, due to the lack of a fixed distant reference, A unfortunately walked in a huge circle. He keeps moving forward, but his front is constantly changing.

　　Returning to drones, what would happen if the assumed heading always has a significant error angle with the actual heading.

　　It will circle in the air, which is "going off course and falling into a vortex":

　　① If it is in a hovering state with small heading error and relatively low speed, then draw a small circle in the air, which shows unstable hovering.

　　② If flying at high speed with a large heading error, it will manifest as a large circle.

　　Note: The reason for appeal reasoning is only a sufficient condition for the result, not a necessary condition.

　　(4) What happened will never be forgotten

　　Abstract: Compass interference is not a transient quantity, it has historical attributes and is related to previous states, so there will always be clean geomagnetic measurements in clean places.

　　Introduction: Why do drones still encounter compass interference at high altitudes, even though there is nothing, it is really puzzling?. In this section, let's talk about these confusing things. Quoting a thought expressed by Mr. Hayao Miyazaki in "Spirited Away": "Please never forget what happened, it's just that you can't remember it.". Try to make everyone understand this principle as simply as possible, but still include some obscure content.

　　Supplementary knowledge 1:

　　The compass is a sensor that can measure the surrounding magnetic field vector (measuring both the magnitude and direction of the magnetic field). That is to say, strictly speaking, the compass should be called a "magnetic compass". Now, it is necessary to understand the concepts of magnetic field, geomagnetism, and northbound direction:

　　① Magnetic field, the superposition of all magnetic fields in a certain place, including: geomagnetism, magnetic fields generated by large surrounding magnetic materials such as reinforced concrete, and magnetic fields generated by surrounding currents.

　　② Geomagnetism, the magnetic field caused by the Earth as a whole, varies in direction and magnitude in different places.

　　③ North direction can be simply understood as pointing towards the North Pole along the ground.

　　Let's discuss the relationship between these concepts again:

　　① Magnetic field is not equal to geomagnetism, but most of the time, geomagnetism is the main component

　　② The direction of geomagnetism is not equal to north, but most of the time, it can be corrected to north by compensating for the magnetic declination angle (the difference between the local magnetic field direction and north).

　　From the perspective of the formula, we can express it as: 1) Magnetic field=geomagnetic+magnetic interference (other magnetic fields) (Equation 2) North direction=magnetic field direction - magnetic field direction deviation caused by magnetic interference - magnetic declination of the geomagnetic field itself. Supplementary knowledge 2:

　　Nowadays, there is a compass on mobile phones, and it usually has a calibration operation, which may not be familiar to those who have used it. The drone also has a compass calibration operation. So what is the compass calibration on this phone and drone really doing? In fact, they use rotating mobile phones to calculate the amount of magnetic field interference. The reading of the compass, subtracting the local magnetic field interference, approximates geomagnetism, referring to equation 1. The magnetic declination of the Earth's magnetic field can be found at a specific location, which can also be subtracted later. This way, the compass can obtain a relatively accurate estimate of the north direction through equation 2.

　　To get to the point, how did the interference from the high-altitude compass come about?

　　There may be many reasons, but let's imagine a scenario like this:

　　A calibrated the compass in place A, obtained an estimate of the interference level of the compass in place A, and saved it in the aircraft chip. Then A flew in place B, and the interference level of the compass in place B was significantly different from that in place A. Unfortunately, drones always subtract the previously calibrated compass interference estimated in ground A, so we will find that the drone's measurement direction is not north because it subtracts an incorrect compass interference.

　　In short:

　　(1) Calibrated in a clean place, flying in areas with interference is problematic.

　　(2) Calibrated in an unclean place, flying in an undisturbed area is also problematic.

　　When calibrating, the interference it receives from the compass will continue to affect future flights, even in open air, and it will have to endure past mistakes Okay. " Model enthusiasts will say, "Then I can only calibrate before each flight, but why is it still not as good as expected?" Then please wait for the next episode: "(5) Information is Fragmented"

　　(5) Information is fragmented

　　Abstract: Fragments of heading information are scattered on different sensors, not just the compass, so the compass has no interference and may not necessarily estimate the heading well.

　　Introduction: Einstein said, "The law of entropy is the most scientific law.". The concept of "entropy" may seem abstract at first glance, but it doesn't matter. We can still gain a vivid understanding of it from a very popular perspective. When teaching perpetual motion machines in high school physics, the law of entropy increase was introduced: "In natural processes, the total chaos (i.e." entropy ") of an isolated system will not decrease." That is to say, the law that Einstein understood as the most scientific law expresses a central idea: "God probably always likes to mess up tables, just like those messy engineers."

　　Entropy was originally used to describe thermophysics, but later it was introduced by Shannon (the father of information theory) to study information, giving rise to the concept of information entropy. Insert a sentence, what is information? Simply put, information is "something more certain than completely random":

　　① Tomorrow is Friday, this is the information;

　　② The price of gold is 80% likely to rebound in the next month, which is information;

　　③ The heading of the drone is 30 degrees northeast, which is information.

　　The more certain the information is, the smaller its information entropy is. However, God always likes to tear intact information into pieces, making people struggle to piece it together.

　　Today, let's talk about: "What kind of fragments did God tear apart the heading information of drones?"

　　The essence of compass interference problem is inaccurate heading estimation. Of course, the sensors that can sense the heading are not known to be the compass. In theory, sensors that change with just one turn of the head contain fragments of heading information. Just follow the philosophical principles in the introduction to understand the problem of heading estimation, and mods will be able to break away from the relatively singular understanding of compass interference, which not only leads to inaccurate heading estimation.

　　The true heading is only known by God. God has torn and crumpled the heading information, scattering it everywhere. These fragments can theoretically be collected by different sensors:

　　The compass picked up crumpled pieces of heading information after God added magnetic interference, noise, geomagnetic declination, and various sensor asymmetry.