Pellicanus - An Open Source GNSS/INS Navigation Development Board

[palpurul]

Hello,

Me and my friend have just pre-launched an open-source IMU board called Pellicanus (because we heard that pelicans are good at navigation ). It's an open-source GNSS/INS(inertial navigation system) board for navigation. I am a PCB designer and I developed the board. My friend specializes in navigation algorithms and me and him wrote a lot of code to implement loosely coupled kalman filter and it's all open-source, so that you can change the code for your own application or uase our code for learning and experimenting with navigation, it's more fun than you think. 

Subscribe to our pre-launch page: https://www.crowdsupply.com/g-fusion/pellicanus

We use Raspberry Pi's RP2040 as our processor. It can be programmed with micropython if you are into it, you can also use C. Micropython can make your development process faster, and potentially making your application's performance slower  , but it's still a nice alternative if you are just trying to get something working.

UPDATE (15.02.2022) -Start
We decided to convert pellicanus to a small SMD module with castellated holes, so that it can be integrated in any project easily.

Here is the layout view of the new SMD module in KiCAD. I tried to make it as tiny as possible so that it doesn't take up much space in the PCB.


Here is some sweet 3D views from KiCAD. I wanted to share renders, but I am not that capable, 3D rendering stuff is not more for me. 


Since it's an SMD module it cannot be used as a single unit, it gotta be on another PCB, and therefore I developed a simple dev board for the Pellicanus.
The dev board includes:

• Type-C USB connector
• 3-pin Speedometer connector
• SDCard connector
• Qiwiic connector
• A few LEDs and a button
• Left-over pins from the RP2040 

Here is a picture of the dev board


UPDATE (15.02.2022) -End

UPDATE (15.02.2022) This is the first version of PELLICANUS which will not make into production


Pellicanus includes accurate and low-drift gyroscope and accelerometer. Pellicanus also includes a magnetometer, barometer and a GNSS module.
Here is a list of the main components:
Gyroscope: IAM-20380
Accelerometer: IAM-20381
Magnetometer: MMC5983MA
Barometer: MS563702BA03-50
However we are going to be switching to IAM-20680 which has both gyro and accelerometer and have the same performance, so performance is going to be the same.

We're experimenting with calibration, our end goal is to facilitate calibration for accelerometer and magnetometer. Here is some data showing the magnetometer output before and after the calibration.

Once the calibration is done, the calibration data is stored in pellicanus for future use. Each Pellicanus board will come with calibration data stored in it. Trust me it really helps with the drift (I will show the drift data befor and after calibration in a later post). We are also preparing documents on how to calibrate the board once you get it. We are working on calibrating gyro, but it's going to take time as calibrating them is more involved than calibrating the accelerometer or magnetometer.

At the end we are going to provide you with the mathmatical model and software for implementing extended Kalman filter, but you can also implement your own algorithms because it's a platform that you can play with precision IMU sensors.

Here is short list for the specifications:
Accelerometer Meas Range: ±2 g, ±4 g, ±8 g, ±16 g
Gyro Meas Range: ±250 dps, ±500 dps, ±1000 dps, ±2000 dps
Mag Meas Range: ±8 gauss
Operation Temperature: -40°C to +80°C
Gyro Bias Instability: 7-8°/hr
Dynamic Heading Accuaricy: ±1°
Dynamic Pitch/Roll Accuaricy: ±0.1°
Position Accuaricy: ±4 meters
Pressure Sensor Meas Range: 300-1200 mbar
GNSS position, speed and heading correction


We are looking for suggestions about what to add to the board.

• What are some of the essential features for an INS board to have that we missed?
• Do you have a cool idea about this project? We like to hear.

More to come later.

[palpurul]

UPDATE!

Subscribe to our pre-launch page: https://www.crowdsupply.com/g-fusion/pellicanus

This is the 3D view of the new version of Pellicanus.


We decided to add a few things:

• Lipo connector along with linear charge, that support a single cell lipo battery
• Extra ROM for logging
• Switched to IAM-20680
• Added terminal for 3-pin speedometer connection, which can be used in navigation algorithm
• Added a qwiic connector right next to USB
• Some cosmetic changes

We're sending the PCB to manufacturing, let me know if you have any ideas to add to the development board.

We're also working on GUI that visualized the raw data from sensors and helps you understand kalman filter and various navigation algorithms better. The GUI is not only going to show you the raw data, it's going to depict orentaiton, lets you fiddle with the parameters in Kalman filter, that way it's going to help you better understand navigation algorithms.

Let us know what you think about it. We need feedback. 

[PlainName]

What's the ballpark cost of this?

[Cerebus]

I'l certainly be keeping my eye on this. Make sure you update us here so that we don't have to keep dropping into crowdsupply to find out when it all actually kicks off.

[palpurul]

What's the ballpark cost of this?

We cannot say that for sure for now, but keep in mind that GNSS and sensors we're using are really cheap, they are precision sensors and they are a bit expensive.

We are advised not to share the price by the crowdsupply team as we are kind of far away from the campaign, but we can promise that it's going to be considerably cheaper tactical grade IMUs (they are in the range of 800-1200$, example: VN-100). We know that VN-100 is not an exact competition to Pellicanus because we obviously focus hobby and education market, and therefore the pricing is going to be suitable for hobbyists and enthusiasts. 

Small correction: (Edit)
Small correction about this VN-100 is not functionally equivalent to Pellicanus as VN-100 is only an AHRS(does not include GNSS). Although you can use Pellicanus as a AHRS you can do a lot more with it because it includes a GNSS, hence it is an INS/GNSS and its functional equivalent is VN-200, it sells for a whopping 2400$  . Obvisouly Pellicanus is going to be a lot more affordable than VN-200.   

I'll post updates here, when we are live with the campaign the price information will be available. I can consult with the community here about the price as we get closer to campaign. In the mean time you can subscribe to mailing list here from this link

[palpurul]

I'l certainly be keeping my eye on this. Make sure you update us here so that we don't have to keep dropping into crowdsupply to find out when it all actually kicks off.

Sure! It's nice to meet people who are interested in what we're doing, it means a lot, thanks! 

Me and my friend will post about the project as we proceed. Right now we are working on GUI that'll be used with Pellicanus, it's going to be great. There are also some work to be done with calibration of of sensors, we are right documenting them, we also like to open-source any calibration related work that we do.

This is my linked profile if you use linkedin you can also follow me there, but it's not necessary because I share in both platforms.

Thanks!

[PlainName]

We are advised not to share the price by the crowdsupply team as we are kind of far away from the campaign

I wonder why. I understand the costs, hence price, will vary as the project progresses, but a good way to turn off potential goodwill is to lead people on for a long time and then have them find they couldn't afford it anyway. Being upfront that it will be in the region of <$100, <$500, <$1000, etc., would mean that your followers are really interested and not just hoping it will pan out in the end.

It's an interesting product which does appeal to me, but I don't know whether it would be something to play with, something to find a project for, or something available if ever a project required that kind of thing.

[palpurul]

We are advised not to share the price by the crowdsupply team as we are kind of far away from the campaign

I wonder why. I understand the costs, hence price, will vary as the project progresses, but a good way to turn off potential goodwill is to lead people on for a long time and then have them find they couldn't afford it anyway. Being upfront that it will be in the region of <$100, <$500, <$1000, etc., would mean that your followers are really interested and not just hoping it will pan out in the end.

I think I understand your point. I can empathize with you, not knowing how much it's going to cost is a big turn-off. I can guarantee that it's going to be <200$ region (I am safe guesstimating ), exact pricing is going to vary depending how many units we think we are going to be selling in the campaign, obviously we can offer lower unit price if we get more attraction. Maybe that's why they don't want us to share the price information in the very early state of the pre-campaign.

It's an interesting product which does appeal to me, but I don't know whether it would be something to play with, something to find a project for, or something available if ever a project required that kind of thing

First of all, thank you for finding our product interesting . That's a good question. I only developed the PCB, in the application side, my friend knows a lot more stuff than I do (his nickname is acsgnn). He will be responding to you about this.

[PlainName]

Thanks. It is much more interesting in the <$200 region

[acsgnn]

We are advised not to share the price by the crowdsupply team as we are kind of far away from the campaign

It's an interesting product which does appeal to me, but I don't know whether it would be something to play with, something to find a project for, or something available if ever a project required that kind of thing.

Hello dunkemhigh

I am developing this project with palpurul. I am a navigation and computer vision engineer.
You can use Pellicanus directly in your unmanned vehicle projects (that’s the first application that comes to mind) or you can use it learn navigation algorithms intuitively with help of our open-source firmware and GUI (more on that later).

It’s worth mentioning that you can incorporate external sensors in addition to what’s already on the board for sensor fusion to make things more interesting and potentially more advanced. For example, you may add a simple radar altimeter, a simple speedometer, or more advanced a visualodometer etc.

In addition, there are many mathematical models for navigation systems. As AHRS (Attitude and Heading Reference System), EKF, Complementary, madgwick etc. Pellicanus is a platform that you can implement all of these algorithm and it comes with AHRS and/or EKF firmware which is open-source.

We are working on GUI that’s going let you change parameters in EKF (extended kalman filter) such as Q and R matrices. That way you can easily play around with the algorithm that we implemented and possibly fine tune the EKF algorithm for your application, and I think that’s also a great way to get intuition on how EKF algorithm Works.

As for Inertial Navigation, Loosely Coupled EKF, Tightly Coupled EKF etc. are used as EKF algorithms. You can develop them all on this board. We will present some articles on these topics.
If you wish, you can directly integrate it into your UAV, UGV, USV project easily as Pellicanus comes with EKF firmware uploaded to it.

In other words, it is a system that you can both play and use in a project.

Let me know if you have more question, we like to hear what you think and we need your feedback.

Thanks!

[PlainName]

Thanks for the extra info. The GUI interface would certainly make a difference when experimenting  and perhaps developing something. Personally, I don't have a crushing need for something like this at the moment, but knowing abilities and where to find them often kickstarts(!) a project.

[palpurul]

Thanks for the extra info. The GUI interface would certainly make a difference when experimenting  and perhaps developing something. Personally, I don't have a crushing need for something like this at the moment, but knowing abilities and where to find them often kickstarts(!) a project.

It's also worth mentioning that we're using RP2040 MCU from Raspberry Pi, it can be programmed with micropython which might be interesting if you're into it.
We used micropython for developing EKF and AHRS algorithms.

[Cerebus]

I think I understand your point. I can emphasize with you,

I think you meant "empathise". 

[palpurul]

I think I understand your point. I can emphasize with you,

I think you meant "empathise". 

Correct, that's what I meant. 

[acsgnn]

Thanks for the extra info. The GUI interface would certainly make a difference when experimenting  and perhaps developing something. Personally, I don't have a crushing need for something like this at the moment, but knowing abilities and where to find them often kickstarts(!) a project.

" alt="" class="bbc_img" />
The default algorithm we use in it. Of course, you can configure it as you wish. You can program it with micropython by generating the best values with the GUI.
You can make a quick configuration with VS code. Or you can use C/C++ and Rust. It's up to you.

[palpurul]

Here are some nicer pictures:

Layout:


3D:

[palpurul]

UPDATE (15.02.2022)
We decided to convert pellicanus to a small SMD module with castellated holes, so that it can be integrated in any project easily.

Here is the layout view of the new SMD module in KiCAD. I tried to make it as tiny as possible so that it doesn't take up much space in the PCB.


Here is some sweet 3D views from KiCAD. I wanted to share renders, but I am not that capable, 3D rendering stuff is not more for me. 


Since it's an SMD module it cannot be used as a single unit, it gotta be on another PCB, and therefore I developed a simple dev board for the Pellicanus.
The dev board includes:

• Type-C USB connector
• 3-pin Speedometer connector
• SDCard connector
• Qiwiic connector
• A few LEDs and a button
• Left-over pins from the RP2040 

Here is a picture of the dev board


Let us know what you think.

We thought it would be better to start the campaign with the SMD module because that way pellicanus can be used both for experimentation/development of navigation algorithms and for integrating into actual projects.

We hope you like the update I've sent the boards to the manufacturer once I get my hands on them and assemble them I will share them immediately 
This is the first castellated hole module that I designed wish me luck 

Subscribe to our pre-launch page: https://www.crowdsupply.com/g-fusion/pellicanus

[PlainName]

Looks like a sensible change to me, and the provision of a dev board is suitable icing 

[palpurul]

Looks like a sensible change to me, and the provision of a dev board is suitable icing 

I'm glad you liked it! 

[palpurul]

New SMD modules have arrived 
This is my first time designing a castellated hole PCB, so I am kind of excited to have these. 



I also designed a base for the module. It's quite simple really, USB connector, SDcard interface, qiwiic connector, speedometer terminal and leftover pins.



I'll post the finished version of it once I'm done with the soldering 

[palpurul]

Here is a picture of Pellicanus SMD. It's a 30mmX30mm castellated hole module.



Lots of flux residue coming from the flux core of the solder wire.
Rate my soldering 

[cgroen]

Here is a picture of Pellicanus SMD. It's a 30mmX30mm castellated hole module.



Lots of flux residue coming from the flux core of the solder wire.
Rate my soldering 

I can't rate your soldering as your cameras ability to focus is a 1 out of 5 

[palpurul]

I can't rate your soldering as your cameras ability to focus is a 1 out of 5 

Well thanks, at least I got a point
My phone's camera is not that great. I tried to focus on it as best as I can, but it didn't work that great 

Here is a more clear version, The lighting is terrible, but you can see my solder work better.