@raghavsharma1214

Refactoring Dataset Generation

Implementing dataset generation checkpoints, in case generation fails, I can just start from where I left off. Also employs synthetic dataset generation.

Basically I’m rewriting the compile_confirmed_dataset and same for negative dataset to include checkpoints and all, because last time when I tried to generate dataasets without this it led to the endpoint giving an error and my whole generation is wasted. Even though it took 12+ hours to generate it

Everthing is practically done! The github is set up, the demo is linked. I just have to recieve the files. Although I did encounter a big bummer… Apparently I don’t get stardust from this.

Trying to get the Speedometer working. Am bruteforcing checking all CAN ID’s to see which one works.

blendering and whatnot

Trying to see how to improve the example image. But I’m also building the github, along with drafting the silkscreen

I’m trying to follow the guide, but improve on eveyrthing, like a better and more extensive readme and whatnot. As per the blender render, I’m taking feedback from my family to see what is too much and what to improve.

Blendering and Rendering

I imported my model into blender for rendering a nice photo so that people have a good idea what I’m making. Probably will re-render with some adjustments, along with turning on denoise and then adding some artificial ISO noise. But hopefully the first ship is coming by soon!

just working on refining things not much up

FreeCAD Slop

Adjusting to FreeCAD is difficult after fusion doesn’t work anymore. After long coercion I am able to create a rough sketch that I can develop further.

fingers crossed, I can hopefully finish this project nice and tidy!

Decoding can messages… I can’t find the can database for this car so Im trying to brute-force finding the codes, but tryna find on the web as well.

The CAN doesnt make sense

The needles are able to sweep, after 2 hours of debugging and brute-forcing CAN messages since I dont have the car, only the instrument cluster. But I have to find the equation that makes it scale number to actual RPM. That will be tough,

Finished the EEPROM Locking

The Audi S5 is known to go into component protection if it is tampered with without the proper ECU TCU gateway modules and whatnot.

So what I did was I forced the EEPROM into read-only. Making it unable to go into component protection, which essentially bricks the whole thing, and instead allowing us to debug and find the signal to use the instrument cluster properly.

Fusion Failed…
Fusion CAD software is genuinely terrible. Doesn’t work on my RTX 4080 even after multiple reinstalls. So I switched to FreeCAD. It is pretty different, but I am getting the hang of it.

I have completed a lot of the CAD work for this project, so we are getting there, to the point where we are done I guess…

Generating datasets…

I tweaked some stuff in the synthetic dataset maker, and added another one that makes synthetic datasets for the negative samples. Generating the datasets right now.

Synthetic dataset is live!

After many struggles, along with downgrading python to 3.11 for compatibility, I was able to generate synthetic datasets. This will significantly streamline the dataset creation process and hopefully speed up the path towards training the model until its nearly flawless

to-do:

• update README to be relevant
• run model training
• optimize model
• run the model to find actual candidates

Trying to fix FUSION. I have ZERO IDEA why fusion isn’t working, just a blank screen. Tried all fixes on the web, but to no avail. Anyone here please help

A MASSIVE dataset

everyone probably knows from my previous devlog that my CNN had massive overfitting. I wanted to fix that.

You know what I did?

I downloaded every single TESS project candidate. Every. Single. One.

Although not all of these candidates are proven to be exoplanets, the algorithms used to detect it market it as very similar to an exoplanet, if not a real one. This will expand my boundaries of my training data, giving me access to ~8k vectors for training.

What I want to do in the future, is to implement an synthetic data generator, Probably using batman. I’ll have to implement things like guassian noise to add some randomenss and whatnot, along with a LOT of RNGs. But we’ll get to that when we get to it.

Wish me luck! My laptop’s about to get fried.

I’ll need to change things around significantly.

Currently, I’m training the CNN with only 2000 vector datasets.

If I really want the CNN to stop overfitting, I’ll have to significantly increase the dataset size.

So here is the todo:

• Use the NASA exoplanet archive to create a list of existing, confirmed exoplanets. Should get over 6000 candidates
• Generate synthetic data using something like batman or someting.

I keep on seeing signs of significant overfitting, unable to be solved. The dataset size must be the problem.

Completed all the routing of the PCB

I realized that routing is a very sketchy task, especially for someone as inexperienced in this field as me. Fortunately, I found a plugin that does the routing for you!

It is called “Autorouting”

Please check it out, its really a life-saver.

Now onto the CAD portion of this project.

Finished wiring the schematics.

I want to use 12 keys so that in the sim-racing game, I can set views, change traction control, brake balance, and all of the intricates.

I am also including one rotary encoder for things like volume control, so I can turn down the volume when my parents come by :)

Setting things up initially.

I don’t really know how kicad works, but that’s fine. I’ll learn.

The struggle right now, is importing everything. I belive I imported the libraries required, but I can’t find the symbols, which is wierd. Cheers!

Changes to my CNN & training.

I tried to reduce overfitting on a small dataset. I increased dropout to 0.5, added L2 regularization, and simplified the architecture by removing a convolutional block. Should improve generalization.

Examining the results of training the CNN model show that there is still significant overfitting. I will need to improve this.

Performance evaluation:

• Training loss drops but val loss remains igher showing overfitting
• Training accuracy reaches very high but val acc lags behind. Shows that model is learning patterns specific to dataset.

To-Do:

• Add time-series data augmentation to increase effective dataset size
  -Reduce model capacity
• Tune regularization
  -Tune optimizer/LR