KI Help for RD DIY Projects

General
1

Maybe you’re in the same situation:
You want to build an RD project and don’t really know where to start.
Your vehicle has neither OBD nor CAN…

In addition to the often difficult task of selecting the right custom DIY hardware (e.g., SBCs/microcontrollers/Arduino,
displays) for Android/Windows or Linux platforms, the question eventually arises: How do you get all of this
programmed without having to delve deeply into the respective programming language/operating system?

This is where Google AI can quickly help and generate all the necessary scripts/Arduino sketches and the RD XML files -
if you ask the right questions.
I asked the AI ​​the following (excerpt):

Create the appropriate scripts for Teensy 3.5, Radxa Rock 5B, and the XML file for RealDash, which can read
the following sensors (Hardware used: Teensy 3.5 connected via USB
Radxa Rock 5B with Bluetooth card
12.3 inch VSDISPLAY (connected via HDMI)):

Connected sensors (just an excerpt…)::

  1. LDR value: Control HDMI display backlight (RealDash XML tag ID = 811)
  2. Read RPM value (square wave signal) (RealDash XML tag ID = 37)
  3. Read digital speed signal G 68 (square wave signal 4000 pulses per km) (RealDash XML tag ID = 64)
  4. Read analog water temperature F76 (RealDash XML tag ID = 14)
  5. Analog value Read oil pressure (RealDash XML Tag ID = 151)
  6. Read digital value of oil pressure switch F22 (RealDash XML Tag ID = 65)
  7. Read analog value of oil temperature (RealDash XML Tag ID = 152)
  8. Read analog value of boost pressure (RealDash XML Tag ID = 83)
  9. Read analog value of charger RPM (RealDash XML Tag ID = 106)
  10. Read analog value of tank content G1 (RealDash XML Tag ID = 223)

The result surprised me greatly.
The AI ​​immediately created all three required files (Teensy Sketch, Rock 5B Python Gateway script, and RD XML file).

It should also be noted that the sensor values, the target IDs, and the frame IDs in the XML file
must all be processed in the same order - important to know…

The AI ​​immediately takes care of the correct order and synchronization after asking “Are the target IDs now sorted and synchronized according to the Teensy and the XML file?” and makes them available.
This way, you get ready-to-use files pretty quickly that only need to be compiled (Teensy) or imported (XML file).

In my opinion, this makes planning and construction considerably easier.

You just have to specify in advance which sensors should be read and also provide them with the appropriate target ID.
You can find the target ID list here:

I would be happy if this helps anyone with problems.

Greetings from Germany

1 Like
2

Thank you for sharing your story. Sometimes AI can nail it directly :slight_smile:

3

There are people who, even if we have that information, will be unable to program and connect Tensy and Pi :sweat_smile::rofl: But thank you very much for the info

4

Hello realdashdev,
Hello DLM,

I’m glad this information might be helpful to others.

Perhaps this could be something for a kind of wiki for Linux users (along with information about the problems with lib dependencies during the RD installation) as a separate topic category in the forum?

The sensors I listed were just a small excerpt from the entire project… the longer you work with the subject, the more ideas for additional functions emerge…

In general, I initially only requested the Teensy sketch for 44 sensors from the AI ​​(the Python scripts for the Rock 5B and the XML file for RD came later).

In addition, there is the dimmer control for the HDMI display using the LDR from the Teensy.

The original idea of ​​controlling the display using PWM via the GPIO of the Rock 5B together with the LDR values ​​from the Teensy was immediately eliminated after receiving the relevant AI advice.

This would have required very extensive interventions in the display controller…

The AI, for example, provided the important tip to clarify whether the HDMI display supports DDC/CI and whether the Rock5B can access the HDMI interface’s I2C bus… Fortunately, both are the case.

This would still make dimming via LDR possible—albeit via a different method…—but requires even more scripting under Armbian Bookworm…

The AI ​​then immediately generated the necessary Python scripts for making changes to the Armbian Bookworm system (with numerous important hints)…

In the meantime, the AI ​​mode suddenly disappeared in Google for no apparent reason (it has only been available in Germany for a few days).
After a brief online search, clearing the browser cache and history brought the AI ​​button back to life…

By the way, compiling the not-so-small Teensy sketch with the AI’s specifications didn’t work right away: The variable declarations were forgotten in some places…

I pointed this out to the AI, which then recreated the sketch.

It compiled without any problems - whether it actually works properly will require some extensive testing…

Once all that’s done, I can move on to the remaining DIY I/O board, including the appropriate protection circuits and the power supply.

Then there are the topics of

  • startup behavior / system optimizations / sleep mode
  • and the missing buttons to enable manual startup / shutdown…
  • Overall, a broad but very interesting field of work…

Greetings from Germany

1 Like