Monthly Archives: October 2020

Intellicam 97 – Designing Autovia – (Part 3)

UI GOALS 🙂 What I hope can be designed at the end (since I am not a designer got one of my designer guys to come up with a GUI)
What we have right now (LOL :P)

So the finished product will probably be called “AUTOVIA” 🙂 so match up with our other stuff Pytavia and Banktavia 🙂

So in this part of the build I hacked up a quick UI to display all the engine readings that I could pick up from the newly developed OBD2 connector for the pre-OBD2 standardized car 😛

It was quite challenging getting the data out in the resolution that I wanted but its now working. There are some small delays as you can see on the UI of the speedometer but its working and just need to slightly tune it so its more real time with no delay.

The other thing I wanted to also get working was streaming the camera from the raspberry PI through a web browser so that the whole UI could be contained in a single web application running of the Pi. In future the goals for this will be to have the video streaming from the Nvidia Jetson nano, so we can get better then 0.8 – 1 frames per second LOL …

Also as part of this phase I got the object detection up and running on the Raspberry PI and its pretty sweet, slow as hell but works as expected using some of the standard models that exist. Will discuss more about what I did here in one of the next 3 posts.

Also to help out anyone who wants to stream video from a PI, below is the generic video streaming code I wrote that can be adapted to be used in any situation where you want to stream video to a website from a camera connected to a Raspberry PI. ENJOY.

Intellicam 97 – OBD2 & Panels – (Part 2)


To get quick access to the code for the OBD2 access you can follow this link and check out the code: I will go into further detail of the code later but for now its there to check out. The goal at the end of this was to learn to use my power tools and ELM327 commands.

I got started on the project by learning a thing or two about how to fix bent, broken and rusted panels. As well as detailing and buffing old and oxidized panels.

To start the project off I bought myself some tools that I needed so I could pull bits and pieces of the car apart as well as sand, cut and buff things out as I needed to. I started off by buying the following

  • Drill kit with 91 bits which included (disk sander, buffer, socket bits and an assortment of drill bit sizes and screws). The drill which I bought is a hammer drill and can go through metal , masonry and wood.
  • Socket wrong and spanner kit so that I could open up door hinges , panels , and help get into the nooks and crannies of the car.
  • Along with the power tools above I bought an assortment of bolts, cutters, paints, primers , buffer pads and pretty much everything else I needed.

Next stage was getting started and putting these power tools into action. So I began sanding down the car and painting. No need to describe this as the images will do a better job of doing it.

Cutting out rust
Fixing deep marks and scuff
Priming the panels
Sanding down the panels

The finished product after I completed the panel beating , sanding and painting of the car is a half decent passenger side look. Still need to buff it down and then add a coat of clear but will do that later on.

Finished passenger side body work

So that’s about it for now

Intellicam 97 -The Intro – (Part 1)

So I have started on a new side project that I think is pretty audacious but will teach me a lot about a multitude of things that I have been wanting to learn and what better way then to use what they taught me at Uni called PBL or problem based learning. This project I would like to call as deemed by my good friend Ahmad is the Intellicam 97. It’s called this because well, I want to make my Camry intelligent.

I have been inspired by all of the new developments in technology specially around AI and autonomous vehicles. I studied AI and machine learning during university in early 2000’s and I never would have guessed how far this technology has progressed now in 2020. Back when I was at uni, it was about movie recommendations, predicting what users would want to buy given they purchased certain previous products and around advertising (it still is today for many companies but for the really innovative ones they have taken this to the next level)

In 2005 I did my masters and did a lot of work with laser range finders , sonar and computer vision in robotics class but back then it was hard. There where no libraries such as what is available today or at least I didn’t know they existed so when we had to build a neural net we wrote the code ourselves, if we wanted to build KNN, we built it ourselves.

It’s different now, I have played with Tensorflow , Scikitlearn, Panda’s , openCV and shit has really progressed. To learn this tech even further and learn a bit of automotive this will be my goal.

Rebuild my Toyota Camry 97 into a fully digital smart car that can see, navigate, warn me of road conditions, take voice instructions and execute them. Its a workhorse and has been in the family for about 12 years. This is what I plan to do:

  • Give the car body a full make over meaning, fix up the paint, rust, suspension and tires. I want to learn a little about panel beating, car detailing and how the steering of the car works.
  • Convert the speedometer and dash items all into digital, Just like a tesla, change all the knobs, switches and dials into a touch screen computer.
  • Install 2 SBC’s on the car , on is the Raspberry pi 4 to handle electronics, servo’s , OBD2 communications and run the general car management app
  • Install a Nidia Jetson Nano to do the computer vision processing so that the car can do better and faster object detection and character recognition.

So what have I done so far ? So far I have completed the following:

  • Fixed some of the really bent panels on my car
  • sanded those panels down and painted them to give the car a nicer smoother looked
  • Fixed some of the oxidized paint using compound rub, drill and buffer
  • Developed the OBD2 communications module from scratch to get data from the OBD2 port (would have used the python built in one but couldn’t cause my car pre-dates the OBD2 standard for Australian cars)
  • Developed some basic object detection using OpenCV and tensor flow
  • Ran the object detection on a raspberry PI and found out that it is (SLOW AF) and hence why will run it on the Nvidia Nano once I purchase it

And that is it so far. On my next post I will go into detail of some of the body work that I have done on my Camry to ensure that it looks classical and original on the outside to complement the high tech look that it will eventually have on the inside.