Hello Gang Just wondering if you know the name of the manufacture that makes the IMUs for FreeFly's gimbals? Any thoughts or suggestions would be greatly appreciated. Thanks! James Aarestad www.BlimpGuy.com
Sensors themselves are off the shelf, however the magic happens in the firmware that Freefly engineered over the years in house..
I struggled with the manual-to be frank. Naively, the modes seem to be: time lapse, dual op, majestic, and target GPS. I have no idea what the default mode is as there is no indicator. The MOVI phone app will tell you if you are leaving a mode such as time lapse, or majestic, but not if you are *in* a mode. The gimbal only tells you if you try to leave a particular mode so it seems like the general settings make a 5th mode, depending on settings you select: 1. aerial or hand held for the gimbal...ect 2. heading assist which has a fixed mount that acts a little like majestic mode in the sense that it initializes to the carbon fiber (handle) frame's coordinates and GPS (that acts a little like target mode in the the sense that it tries to hold a bearing) and finally compass, which has its own problems. My goal was to shoot a scene of a mountain peak while driving a windy road and stay locked on the peak. It has been easier said than done. I was using the SDK commands for feedback to the MOVI to choose a bearing relative to the initial bearing (with GPS heading assist on). It worked ok until I hit really windy roads with washboard grooves in the pavement. The compass seems to get angry under hand accelerations and loses a good compass bearing. So then you start to wonder about the IMU. Majestic mode really cares about the frame's coordinate system, so fixed mount wasn't useful for my application. I don't care about the frame coordinates, but the earth's are useful. GPS heading assist was decent at pointing in one direction but the drift made me wonder if they are incorporating Kalman filters to tie the IMU to their GPS. The drift got bad enough even with a good GPS signal that the camera drifted 180 degrees away, but continued to report through the SDK data stream that the bearing was held (this was pulled out using the SDK and the quaternions. I verified this data stream with the internal tracking on the SD card (this yaw data matches the SDK quaternion data stream). In any case, I took it apart to learn more about the IMU...seems to be a TDK InvenSense MPU-6000 Accelerometer, Gyroscope, 3 Axis Sensor I²C, SPI Output...see https://store.invensense.com/datasheets/invensense/MPU-6050_DataSheet_V3%204.pdf The zensor I²C said that it can be coupled to a magnetometer, but I do not see an explicit magnetometer chip for compass bearing...so I am struggling with maintaining a accurate YAW (Pan) coordinate system. This should answer your question. Hoping someone know the magic setting parameter so I don't have to mountain climb to the peak of that mountain to use target GPS so as to maintain pointing at the mountain peak.
THANK YOU ROBERT!!! Thats exactly what I was looking for! Do you know who makes their gimbal controller (GCU)?
Hi James, it is probably not what you pictured. I should also correct myself, from above...having learned more. The generic mode, I believe is the MIMIC mode: not majestic, not timelapse, not dual op, and not target GPS. If not in one of these modes, the back of the gimbal reports "Mimic" Regarding your question: By symmetry, it looks like there maybe driver chips for the three wire stepper motors. In each module there is an 32 bit ARM chip, this will probably function as the brains of the gimbal. Nothing else appears to be control logic. There are some power N channel MOSFETS, which was good to see. At least they are using an efficient switching transistor...should save power. The ARM processor is: STM32F746xx, https://static6.arrow.com/aropdfcon...3eac0cec0f1249/6467314819840706dm00166116.pdf I wouldn't have gone to the extent of looking into this, but wanted to learn more technical stuff. A lot of confusion could be cleared up with a good matrix/table of which settings impact which modes. Not really wanting to reverse engineer it as understand what I can do to make it work for what I want it to do. BTW, the "no heading assist" seems to work the best at maintaining an angle. I found a fair amount of drift in the horizontal pointing angle when I tried GPS heading assist, compass, and fixed mount. Even through it may say compass, GPS, and lock good. It would drift about 90 degrees after driving down a windy road. To be fair, I didn't go through the compass calibration, because I didn't want to take it out of my vehicle mounted test rig *every time* I took it out. Maybe it would do better after a compass calibration? I would be curious if the Movi XL requires rotating the whole assembly for calibration and how that is practical for vehicle shots.
*Quick response, apologies for typos, etc Freefly makes the GCU. Again, most of the magic is on software and not necessarily the hardware components The default mode is called Majestic. Majestic is a patented technology that allows MōVI to be controlled intuitively with a handlebar. Before this technology, gimbals were controlled by remote inputs. When there is an external Rate or Absolute or MIMIC command to MōVI, this is called Dual Op. Its a generic term. Timelapse is a motion control mode. You can imagine that being an "absolute" command, where MōVI is controlling itself. Target mode is similar to this as well, however, note that it is an experimental feature. "GPS" mode can be confusing if you don't clarify what you mean. There used to be a GPS heading assist mode, which used to improve the stabilization in high acceleration cases (such as taking a sharp turn with a car), however this mode is removed in Blackjack software update and now MōVI does all this behind the scenes. Heading assist still has the other two options: Fixed Mount and Compass. These are not recommended unless you have a very specific use case. Fixed mount used to be used when setting up MōVI on a tripod for long timelapses. It engages pan encoder to be a reliable source for attitude estimation, does eliminate slow IMU drifts. However this assumes the "base is stable". So if you move the base (or handle), it may look like "majestic" even when you are in dual op. Note that this setting is automatic in Blackjack. MōVI automatically knows when the base is stable, and automatically switches to fixed mount behind the scenes. Lastly, the compass is useful for absolute heading if you have a special use case. However if you don't have special needs, then it is not recommended since compass gets affected by ferromagnetic materials and can do more harm than benefit if not paid attention. There is no "aerial" mode. There is a "gimbal application" setting that changes MōVI to be in a special mode when not being commanded remote commands. This allows for easy take-off and landing, so that ALTA doesn't spin on top of MōVI. This is a feature, more than a "mode". "My goal was to shoot a scene of a mountain peak while driving a windy road and stay locked on the peak. It has been easier said than done." I am not sure if there is a clean way to do this with what the current public API exposes. DM me and I might be able to help when I get some time.. Note that for what you want, you need MōVI to know its absolute heading. For this you will need compass heading assist and a good calibration.