IMU Sensors 101 – Pros & Cons of Sensor Types
One of the most common questions we get at Inertial Sense throughout the industry is what am I looking for and what’s the difference between IMU sensors and their options, including AHRS, Kalman filters, and so on.
Watch this video below as Carson talks you through each individual module and the difference between them and some of the features that they include.
IMU Sensors: What is an IMU?
IMU, which stands for inertial measurement unit, is the core of our sensor products. This is a sensor with several different sensors inside. It has accelerometers, gyros, magnetometers, and barometers. When you purchase an IMU, you get raw data from the calibrated sensors. When you plug an IMU into the inertial sense firmware, the firmware will tell you what the accelerometer, gyro, magnetometer, and barometer are doing while informing you of the data being seen.
IMU Sensors: What is AHRS?
The AHRS, attitude heading reference system, contains the same components as IMU and outputs the same raw data. What makes the AHRS special is the Kalman filter. A Kalman filter makes assumptions consisting of a series of algorithms that fuse all data together and outputs what is called the roll, pitch, and yaw. Roll can be thought of as a barrel roll. Pitch is the forward and backward tilt of your device. Yaw is given by the magnetometer and is your compass direction that the application is facing.
How a Kalman Filter Helps
A Kalman filter is a series of algorithms. It takes every single sensor inside the IMU, including the GPS, and fuses all the information together. The Kalman filter uses a wheel odometry sensor to fuse that data with the GPS as well as the data from the magnetometer and gives a more accurate idea based on where the robot was to where it is going because of all the different sensor fusions.
IMU Sensors: How is INS Different?
The highest grade or the highest quality version sensor we offer is the INS. The INS is a combination of all the sensors. INS, which stands for inertial navigation system, combines the AHRS and the IMU. The system consists of a bunch of different tasks being accomplished, taking the IMU data and using a Kalman filter to fuse all the IMU data together, then fuses all the fused data from the AHRS together with the GPS, giving a higher precision.
How Do I Decide if IMU, AHRS, or INS is Best?
The customer that purchases an IMU is a customer with unlimited resources. These are your multi-million-dollar companies that will spend millions on the product being developed because they have all the resources they need to make their own Kalman filter and build the algorithm themselves.
The person purchasing the AHRS does not need GPS fusion, or they have similar experience making these algorithms, but they want to do their own GPS fusion. These will still be larger companies with a significant number of resources.
The INS is for the person who wants the machine in the market as fast as possible. They do not want to spend their time and resources making their own filter. This is our most popular product because autonomy is up and coming. It is something that is happening today. And if you want to keep up with the market and the trend, you need to act now and click here!
One of the most common questions that we get throughout the industry is what am I looking for and what’s the difference between the sensors that you provide. During this video, I’m going to talk you through each individual module and the difference between them and some of the features that they include. The first one that we’re going to start with is the IMU because it’s the most basic. It is the core to the rest of our sensor products.
0:32 What is an IMU?
The first sensor that we’re going to go over is the IMU, or the inertial measurement unit. This is the most basic of the three, and the most important thing about the IMU is that each one of these modules contains IMUs inside of it.
Alright, so the IMU is a sensor that has several different types of sensors inside of them. It has accelerometers, it has gyros, it has magnetometers, and it has barometers. When you purchase an IMU, what you’re getting is the raw data from the calibrated sensors. So when you plug an IMU into the inertial sense firmware, the firmware is going to tell you what the accelerometer is doing, and the data that it’s seeing. It’s going to tell you what the gyro is doing, and the data that the gyro logs. It’s going to do the same with the magnetometer and output that raw data as well as the output from the barometer. In addition to these individual sensors, the IMU is GPS-aided. So you are also going to receive the raw data that is being given from the GPS. when you purchase an IMU, that is the data that you’re going to be provided with. What makes these other sensors better than the IMU is that they still have all this IMU data inside, but there’s something we call a Kalman filter, which fuses all this data together, and outputs something easier for customers and engineers to understand, which leads us to the AHRS, or the attitude heading reference system.
2:12 Breakdown of AHRS
The AHRSs has all of these same components inside of it, and it outputs the same raw data. What’s special about the AHRS is the Kalman filter. What a Kalman filter does is a Kalman filter makes assumptions and it’s a series of algorithms that fuse all this data together and output something called the roll, pitch, and yaw.
If you’re unfamiliar with these terms, roll I like to consider like doing a barrel roll. I like to consider the pitch the forward and backward tilt of your device. And the yaw is given by the magnetometer. So it’s basically your compass direction that the application is facing.
3:04 What Exactly is a Kalman Filter?
One of the terms that I use while explaining the AHRS module was a Kalman filter. Let’s go into a little more detail of what a Kalman filter is. A Kalman filter like I said is a series of algorithms. So you have the IMU data, that’s the raw data that is being output by the calibrated sensors. You have a robot. We’ve got a robot here. Let’s give him an antenna. This is a little rover that drives along sidewalks. This is a delivery robot. And this is delivering food for a local restaurant called Woody’s. Woody’s robot delivers barbeque. And it runs along a sidewalk to make its deliveries. Let’s say roughly this sidewalk is about three feet. The Kalman filter is going to take this IMU data, and it’s going to fuse it with the GPS. GPS is only accurate to two and half to three meters. So what does that say if you’re just using GPS on a three-foot sidewalk. Say this is your map right here, and these are crossroads to a street. Your robot technically should be right here on the side of the road. But because of the error that comes with GPS, your robot could be anywhere inside of this circle. This is an issue because if you’re on a three-foot wide sidewalk, and the GPS is only reading to two and a half meters in accuracy, you will never know if your robot is actually on the sidewalk or if it’s somewhere close to it. Okay, if you come too far off the sidewalk, your robot could potentially get run over or hit by a car. A Kalman filter takes every single one of the sensors inside of the IMU, including the GPS, and it’s able to fuse all of that information together. So say your robot knows exactly where it is when you start. The Kalman filter is able to use a wheel odometry sensor to fuse that data with the GPS as well as the data from the magnetometer, and it’s able to give you a more accurate idea based on where that robot was to where it’s going because of all of the different sensor fusions going on. So now if the robot has the magnetometer tell it it’s now no longer going south but going east, you know that it is moving in the wrong direction, and the Kalman filter can help you correct that issue inside of the computer, inside of the robot itself.
When understanding a Kalman filter, it’s very important that we understand that there are different grades of Kalman filter that can be provided. In the case of an AHRS, or an attitude heading reference system, although that device is GPS-aided, the Kalman filter is going to be limited to just fusing the data that you receive from the IMU itself. So the GPS, you’re still going to receive raw GPS coordinates, but unfortunately with the Kalman filter involved with an attitude heading reference system, the fusion is going to stop there with roll pitch and yaw. If you need your GPS data fused into the mix, we recommend that you go with an INS, an inertial navigation system. And that’s because it contains what we call an extended Kalman filter.
6:41 Is INS For Me?
The third sensor product that we offer, the highest grade, or the highest quality version is the INS. The INS is a combination of all the sensors down the line. It’s a combination of the AHRS as well as the IMU. INS stands for inertial navigation system. The key word right there is inside the s. This is a system. A system basically consists of a whole bunch of different tasks being accomplished. So it’s taking all the IMU data, it’s using a Kalman filter to fuse all that IMU together, and then the final step that it’s taking is it’s fusing all of that fused data from the AHRS together with the GPS. This gives you higher precision.
One of the best examples that I can think of is our INS onboard adrone. Say we’re using this drone for precision agriculture. It’s doing a survey of some kind. So if you’re using an AHRS, it’s just going to give you the roll, pitch, and yaw of the device. Once that data becomes fused with the GPS, the GPS gets a higher update rate up to a kilohertz, and it’s able to update as it moves and this data is going to continue fusing. If you lose GPS for up to five seconds, you’re not going to lose its path because the sensor fusion is going to take over. And as long as within those five seconds we’re able to get the GPS back, you don’t lose the valuable data that you were using.
So to help you better understand the function of an INS, it takes all the data from the AHRS. so the roll, the pitch, and the yaw to help the system understand the orientation of the drone to determine where it’s going to be. In addition to that it fuses with the GPS. GPS’s have what we call update rates. So as the flight, as this drone continues to fly, it’s going to basically be dotting out its path up to the kilohertz. And so every little motion that this drone goes through, position wise, is going to follow that and it’s going to help the filter in the output in the software be more precise and more accurate than that of the AHRS or the IMU. Because instead of getting raw GPS coordinates, you’re getting a fused position to go along with the motion data that’s coming from the inertial sensors.
So compare this level of accuracy of an INS to that of just your IMU. It’s still going to have all of this data, along with the raw GPS coordinates, but that’s only going to update at five to ten hertz. That means that there is all this space for error wehr you don’t know where your device is versus a kilohertz that has a significantly higher amount of update rate.
So just in summary, you have the fused sensor data, where the data, the inertial data is fused with the data of the GPS that outputs at a higher rate to the data over here of just the raw GPS coordinates. There’s significantly more room for error in between the boxes that are spaced out farther apart. So if you have these gaps in your data, it leaves room for error.
10:27 Inertial Sensor Application
Each individual sensor has a wide variety of use case options. These use cases are actually very similar to the sense that it really depends on the type of company that is purchasing these sensors you can use an IMU in place of an INS. If you do, you need to be prepared to build your own Kalman filter that’s going to fuse it’s sensor data to make it an AHRS, and also fuse the data from the GPS. so an IMU has a wide range of use cases. You can use it in autonomous cars for example. You can use it in using a quadcopter. You can use it on a marine boat of some kind or an autonomous ship. If you are to use an IMU, be prepared to use your resources which can, which may be significant to developing your own filter.
The same goes for an AHRS. a lot of people have purchased an AHRS from us for use cases, and a lot of them don’t need the GPS fusion. One of my personal favorites is there was a cave that was being explored by a university down in Peru. They had a drone that they were sending in to a mine shaft, and they had no need for GPS. Therefore, the AHRS was the better option for them because they didn’t need to create their own filter or algorithm to fuse the GPS data, but at the same time, they were able to accomplish everything they needed to just by having the roll and pitch and yaw.
If you’re using an INS, which is a lot of startups. A lot of startups that work with us are companies that frankly just don’t want to utilize their resources to build their own filters. It takes a long time to make the algorithms you need for an INS to perform the way that you’d like it to. So if you don’t want to make your own filters, and you want something that comes out of the box ready to start swinging, the INS is the best option. We’ve got use cases in quadcopters. Our sensor is in cars, our sensors is inside marine applications of many different types, and defense and aerospace is using this kind of sensor in a lot of different ways because they want to be building their toys. They want to be making things. They don’t want to be dealing with software algorithms to make the GPS function better.
13:14 How Do I Decide if IMU, AHRS or INS is Best?
The type of customer that would purchase an IMU is a customer with unlimited resources. This is going to be your multi-million dollar company or a company who is going to spend millions possibly billions on the product that they’re developing because they’re going to have all the resources they need to make their own Kalman filter. They’re not going to need someone else’s outsourced algorithms. They’re just going to do it themselves.
The type of person that is purchasing an AHRS does not need GPS fusion or they have similar experience making these algorithms, but they basically want to do their own GPS fusion. These are still going to be larger companies with a significant amount of resources.
The INS is for the person who wants to take the machine to market as fast as possible. They don’t want to spend their time and resources making their own filter. Frankly they want an out-of the-box solution that they’re able to plug in and get working as fast as possible. And this is our most popular product because autonomy is up and coming. It is something that is happening today. And if you want to keep up with the market and the trend, you need to act now.