search

UVicFH / Home

UVic Formula Hybrid Documentation Project Management
1 stars 0 forks source link

Telemetry #32

Open chadjmccolm opened 6 years ago

chadjmccolm commented 6 years ago

Context

We want to send the car's data to a server and display it on a webpage to reduce the amount of information clogging up the driver's display and allow for easier data logging.

We have previously purchased parts for this: https://www.ubnt.com/airmax/rocketm/ https://www.ubnt.com/airmax/nanostationm/ https://www.ubnt.com/airmax/airmax-omni-antenna/

The data will be sent via the MQTT protocol to a server created by @BrendonEarl.

The server is powered by NodeJS. The data will likely be sent by an Arduino running an MQTT client like this one.

The flow will be as shown here: img_20171115_201532

Outcomes

Constraints

We should use the technology we've already purchased.

MrMikeGrant commented 6 years ago

With regards to testing the telemetry system, I have been making progress with a HIL simulation. Essentially the MABX will be wired up to a dSPACE simulation platform and receive signals in real time that it would normally see on the car. If we wire up the telemetry system as shown above, we could test and validate its implementation without requiring the car to be running. ETA is mid January.

chadjmccolm commented 6 years ago

Arduino in the above schematic will be done instead with Raspberry Pi which will also power the dashboard. What is displayed on the dash is up for debate but I think what we really need is:

Everything will be sent to the MQTT server using a python library and the dash will also be powered the same Python program.

This also opens up the possibility of sending voice data back and forth for the driver to communicate with the pit. This year, we likely won't have time to do this but it's a good future plan.

chadjmccolm commented 6 years ago

Here is my idea for the dash. Working on building the GUI currently. I've decided to use Python because it has great libraries for MQTT and CAN communication.

photoscan

So far the pre-reqs for getting this running on the RPi are:

The program will run in fullscreen by default and the Pi should be configured to run this program on boot.

pyside can be installed with pip with the commend pip install pyside

chadjmccolm commented 6 years ago

I wrote a summary of the telemetry system for the Whiton Fund which is available here

Good resource for future reference.

chadjmccolm commented 6 years ago

A fuel and charge level indicator will be placed on the UI on either side of the current information or maybe below it. Haven't decided yet. The libraries will be switched to PyQT and python 3 to correspond with the CAN library I want to use, python-can. The QML file controlling the GUI will likely remain the same, the only changes will be to the python application itself.

bndnio commented 6 years ago

@chadjmccolm with regards to the display: I'm not sure if you were previously aware of this, but part of the motivation for the telemetry system was so that we could simply the dash and no longer use the display as it was hard to see on bright days. While it does look super sexy, it wasn't super functional when racing. We thought a bright LED array would be more noticeable as a tach. Grayson previously had a design, which should be on grabCAD or the drive somewhere, for a new dashboard. Not that you guys should definitely drop the screen, but it may be work exploring

tmurias commented 6 years ago

@BrendonEarl The dash design Grayson and I worked on last year is in the Electrical/2016-2017/Dash folder in the drive. I was going to change some parts of it this season but put it on the back burner since I'm spending all of my time on the AMS, and since Chad wants to combine the dash with telemetry I figured we'd just abandon it. It's still there (and pretty much complete) if we change our minds, but it might be nice having the dash and telemetry combined. As long as the new LCD will be bright enough and we only display the really important stuff, it should be fine.

chadjmccolm commented 6 years ago

The plan is to ditch the dash next year and improve it this year. I'm almost done recreating the UI using the PyQT library. Just gotta add fuel and battery level and then I'm Gucci. I've also switched all the graphics to SVGs and did all the positioning relational so as long as the aspect ratio is somewhere near 16:9 it should look great regardless of screen size and pixel density. I'm also probably going to make a shroud for it of sorts to block out light.

Design note: I've changed the tach from a linear with torque curve to a radial because demoing the operation before showed me that our gear ratio is too close for the lowered resolution of a linear tach and the larger sweep of the radial tach is the best idea.

chadjmccolm commented 6 years ago

Version 2.1 of Teledash complete. Preview available here: https://drive.google.com/file/d/1Qh-QPWryCmi-VmfAidZ1QDS9iwBabsGg/view?usp=sharing

Fuel and battery has been added with a backdrop so you can see where 100% after they've started going down on suggestion of @smithbeuvic

chadjmccolm commented 6 years ago

Mosquitto is going to be our MQTT broker. If it runs on a windows system it needs to be installed with some dependencies. When installing them, INSTALL OPENSSL < V1.1 OTHERWISE IT WON'T WORK. Learned that the hard way.

Paho MQTT will be the powering the send to the server. It can be installed via pip: pip install paho-mqtt

chadjmccolm commented 6 years ago

Installation instructions for dash code on windows system:

Now the code can be run. (this will likely be modified soon)

Clebeuf commented 6 years ago

@chadjmccolm: Reminder about the list of data points for the telemetry dashboard. A general list would be great for now, but eventually it would be great to sit down and create a prioritized list of data needed for each use case for the telemetry dashboard.

Once we have an idea of the data (and it's) general structure we can get started on the front end of the telemetry dashboard. Then hopefully when the car is ready to transmit it, we can hook it up pretty easily with a little tweaking.

chadjmccolm commented 6 years ago

A whole lot of dependencies downloaded later, the program is now working on the Raspberry Pi 2 running Ubuntu Mate just the same as it works on Windows. I'm going to remove a bunch of unnecessary programs from the OS and then create an img for reference.

chadjmccolm commented 6 years ago

Here's how to set up mosquitto on windows so that we can run code that simulates signals coming in from the Pi.

  1. Download mosquitto.
  2. Install Mosquitto. Don't worry about the dependencies right now. Installation may appear to not complete or fail.
  3. Download and install Win32 OpenSSL v1.0.2n Light. Install the dlls to the OpenSSL/bin folder. We need to grab them later.
  4. Download Pthreadvc2.dll.
  5. Copy libeay32.dll ssleay32.dll from the OpenSSL32(/bin) folder and pthreadVC2.dll into the mosquitto installation directory.
  6. If you don't already have Visual C++ Redistributable Packages for Visual Studio 2015 you will need to download and install that as well or whatever year's visual studio was used to compile on their website.
  7. Reinstall mosquitto so that it can use those DLLs to make itself a service.

At this point, I would go into services (run services.msc) and change the start type of the mosquitto service to manual so that you can start and stop it as required.

If you cant start it as a service you can always run it by starting up a command prompt, navigating to the mosquitto directory, and then running the mosquitto program by name. In my prompt, it reads: C:\Program Files (x86)\mosquitto>mosquitto.

chadjmccolm commented 6 years ago

@Clebeuf here is the data that I can supply the telemetry dashboard as far as I know. Not sure what we're getting from the accumulator yet because we don't have one yet.

https://docs.google.com/document/d/1b_Rnik_16HKZxufBX0N3VIh2z4FQfbZCLo3LWnhIgOw/edit?usp=sharing

Stored in our Drive in Drive\Software\2017-2018\Signal Hierarchy.gdoc