Pairing super beacons leads to inaccuracy

[vdsander]

Hello,


My setup is :
- NIA
- 4 stationary beacons
- 2 mobile beacons
- HW 4.9
- noisy environment (server rack and acoustic measurements)
The 4 stationary beacons have the same frequencies (31 kHz)
The 2 mobile beacons each have a different frequency (25 and 27 kHz)

It is a project for my master's thesis in which I use a robot to make measurements throughout a room and Marvelmind as navigation system.

The 2 mobile beacons have good tracking in Dashboard and via Python when they are not paired.
As soon as they are paired via UART or radio (with increased distance), tracking becomes less stable and when enabling fusion of IMU, it is no longer usable.

Do I need a license for pairing or is the license only for fusing IMU?


Thanks,
Sander

[admin]

Hello Sander,

If you have Beacons HW v4.9, they can work only at their native frequency for both ultrasound transmission and reception. Therefore, if your stationary beacons are 31kHz, your mobile beacons must be 31kHz only. Super-Beacons have digital filters. They could receive 25kHz and 28kHz or 37kHz (by the way, we don't have 27kHz beacons), but Beacons HW v4.9 have analog filters and can only receive their native frequency. Check the label on the bottom of the beacon to see its native frequency.

Also, remember that you can't set the frequency of the beacon arbitrarily - use only its native frequency: https://youtu.be/JmVjpypEc04.

Recommendation:
- If you already have 4 x Beacons HW v4.9, you can build a 2D submap and use the remaining 2 beacons as mobile one. You will have something like this: https://youtu.be/aBWUALT3WTQ?t=97
- Get Super-Beacons as mobile. They have much sharper filters - better for noisy environments. You can reuse your existing Beacons HW v4.9 as mobile
- It is recommended to have Modem v5.1 as more capable. We stopped adding new features to Modem v4.9 about 2-3 years ago

BR,
Maxim

[vdsander]

Hello, thank you for your answer.

A few things I should specify:

The beacons, stationary and mobile, are super-beacons. Only the modem is HW 4.9.

The set is a combination of 4 stationary beacons with native frequency 25 kHz and 2 mobile beacons with native frequency 31 kHz.
I have chosen 31 kHz as the working frequency since this got me the best results. Since the stationary beacons are also super-beacons, this should not be an issue right?

To be able to get an idea of the accuracy in the environment I have (open room with server rack), I measured the static position of both hedges in three parts of the room separately. The measurements took place by setting the hedge, collecting 100 position updates, moving it over 1 meter and again collecting 100 updates. I moved the hedges along the x-axis first and then along the y-axis second (see attachment).

The weird thing is that there is a difference in the accuracy between the two hedges. In the left part of the room (location of server rack), the accuracy is lower for both hedges, as expected. Both in the left and right part of the room, the first beacon outperforms the second beacon. Only in the middle of the room, the second beacon performs better than the first one. What is really strange to me, is that in the right part of the room (farthest away from the server rack) the difference between the hedges is immense. Beacon 2 performs 20 % worse than beacon 1, in the same conditions. In the other parts of the room, the difference is limited to 3 %.

I was wondering if you could provide me with some possible explanations as to how the performance of the beacons can be so different, eventhough they are the same model (super beacons), have the same native frequency (31 kHz) and were in the same test setting.

Thanks in advance.

Kind regards,
Sander

[admin]

Hello, thank you for your answer.

A few things I should specify:

The beacons, stationary and mobile, are super-beacons. Only the modem is HW 4.9.

It is possible to use this configuration. But, of course, it is highly advisable to have a newer modem because we stopped adding new features to v4.9 series a few years ago. Thus, the modem may be limiting the whole system at some point.

The set is a combination of 4 stationary beacons with native frequency 25 kHz and 2 mobile beacons with native frequency 31 kHz.
I have chosen 31 kHz as the working frequency since this got me the best results. Since the stationary beacons are also super-beacons, this should not be an issue right?

Note a problem for NIA but would be a problem with IA.

To be able to get an idea of the accuracy in the environment I have (open room with server rack), I measured the static position of both hedges in three parts of the room separately. The measurements took place by setting the hedge, collecting 100 position updates, moving it over 1 meter and again collecting 100 updates. I moved the hedges along the x-axis first and then along the y-axis second (see attachment).

The weird thing is that there is a difference in the accuracy between the two hedges. In the left part of the room (location of server rack), the accuracy is lower for both hedges, as expected. Both in the left and right part of the room, the first beacon outperforms the second beacon. Only in the middle of the room, the second beacon performs better than the first one. What is really strange to me, is that in the right part of the room (farthest away from the server rack) the difference between the hedges is immense. Beacon 2 performs 20 % worse than beacon 1, in the same conditions. In the other parts of the room, the difference is limited to 3 %.

I was wondering if you could provide me with some possible explanations as to how the performance of the beacons can be so different, eventhough they are the same model (super beacons), have the same native frequency (31 kHz) and were in the same test setting.

We would need a bit more data/photos of how exactly everything was measured and what exactly "20% worse" would mean.

Here is an example of the tests we did some years back when a similar question arose:
- https://youtu.be/qYJzxq8YC1A
- https://marvelmind.com/pics/marvelmind_ ... cs_v02.xls



Thanks in advance.

Kind regards,
Sander