currently the whole world is affected by SARS-CoV-2 and the resulting measures and so are we. We planned to shoot and edit our first video right around now, but these plans fell through as announced on the Studio blog. To make the most of it, we decided to start with the development of the Inertial Navigation System.
Goals, Goals, Goals
One day the INS will be one of the core systems onboard the Poseidon rocket. Our goal is to develop the system with the following capabilities:
- Track the Orientation of the rocket during the flight
- Transform vectors from the local to the global coordinate system (and vise versa)
- Estimate the position of the rocket relative to liftoff (maybe with additional barometric sensor)
Currently our focus is on the first and second goal, because with these things sorted out, we can start the development of the stability system. We need reliable orientation data to calculate the difference between the rockets rotation and the targeted rotation. Furthermore vectors related to the TVC system needs to be transformed between the different frames to better estimate the effect on the rotation of the rocket for a specific TVC movement.
The basics for these first goals are more or less completed. As you can see in the demo video down below we achieved a working orientation tracking. Because we have the constraint that the vehicle is moving, the acceleration measurements can not be used to calculate the orientation during the flight. So at this point in the development only the gyro data is integrated over time in small intervals, which results in the current orientation.
The integrated gyro data enables us only to calculate the orientation relative to the rotation of the vehicle in the beginning. But how can you calculate that initial offset? To compute this orientation we used the assumption that the vehicle is not moving in the beginning. With that we can calculate the orientation which rotates our acceleration measurements to point downwards (as gravity normally does).
We chose to solely rely on Quaternions to represent orientation, because this is the most elegant way with the least restrictions. We will discuss this topic in a broader fashion in an upcoming video.
Gyro measurements are subject to drift. That means that over time they will start to falsely indicate the angular rate. To counter this effect it’s our plan to implement a Kalman filter for the gyro data to minimize this drift. When it is implemented and working we will start with the development of the stability system as mentioned earlier.
We also want to do an in-depth video about all of the topics related to our INS-Development. It will be released on our YouTube channel and on the Studio subsite. The scripting is well on its way and this are the topics we want to talk about:
- Pros and Cons of different mathematical representation of rotation (Euler Angles, Rotation Matrices, Quaternions)
- The basic math behind Quaternions (with the help of 2D-Rotation with Complex Numbers)
- Our Implementation of gyro data integration
- The basic math behind Kalman filters
- Our implementation of a Kalman filter
- Calculation of the initial orientation with acceleration data
- Conclusion and a small outlook
If you are interested in a particular topic in that regard, let us know it in the comments. We will make sure to include it in the video. This video is second in line after our introduction video, so stay tuned. You can view the before mentioned demo video right after this paragraphs. That is it for today! We are thankful for every feedback and this is just the start. See you soon. Bye!