Hello Everyone, I've been wanted to go in robotics with main focus on robot learning and perception. And for sometime I've already have started learning and also trying to get a roadmap that have everything. And since people tells there are millions of ways to do something you want to do when i asked in this subreddit. I started to search about the skills i need and then constructed a roadmap and then gave it to LLMs such as Claude and Qwen and then got this final roadmap and wants to know if this is good to follow or not:

https://docs.google.com/document/d/1YsCt6O4s-PvEzWACL_PwVSyg85zf3ZcFW8kIOKCR72U/edit?usp=sharing

Note: I modify roadmap if i see anything mission when I go to the steps. So, It does not need to be perfect. Just good to follow.

So, The current version is written by AI but not fully by AI all of the steps I have searched on my own and written and and then to fill gaps and to write it better i have used AI as many people do advice. I already have knowledge in ML and NLP. So, I did not focus in basic programming languages and machine learning. Just going to start from computer vision in ML part. And here I want someone human to review this and tell me if its good to follow or not.

One thing also I'm currently not in college and will go to college this year. But It won't be degree in robotics directly but in computer science or something similar and that is why I need roadmap for robotics.

Thanks.

Hey everyone,

I'm self-teaching robotics with a focus on perception and robot learning. I learn best by building, so I recently started learning ROS2 and Gazebo. I took a rover model, assembled it into a URDF, and tried to calculate the mass and inertia for the components manually using volume and material density.

Here is the repo: https://github.com/introlix/robo_car

Note: you can ignore the esp_control folder. I originally started this for a physical ESP32 car but moved to simulation so I could learn Gazebo physics and sensor integration before touching real hardware.

Since I'm doing this alone, I'm relying a lot on trial and error and AI tools to help me. But I know AI hallucinates.

If anyone here has experience with Gazebo/URDF, I’d really appreciate it if you could take a quick look at my URDF. I mainly want to know if my mass/inertia numbers look realistic, or if I messed up the math and my robot.

Also, if anyone is also an student then we could learn together. I'm not looking to pair-program on a call, just someone to do reviews on GitHub, share resources, and maybe give each other small weekly challenges.

A bit about my background: while I'm relatively new to ROS2, I have some ML background. I've built neural networks from scratch in NumPy and actually implemented LLM architectures (like Gemma and Qwen) from scratch just by reading their papers and loading the weights. That is the reason I'm interested in perception and robot learning.

Let me know if you're open to reviewing the code or if you want to team up. Thanks!

Note: Please read the full post before replying. This is NOT just another low-effort LLM wrapper. It has built in scrapers, databases, and is meant for different kinds of ML tools, not just text generators.

Hey everyone,

Over the past few months, I’ve been doing a lot of deep research and found myself needing a bunch of different tools. Honestly, I absolutely hate how mainstream cloud tools like ChatGPT or Gemini handle research. They hallucinate, they make stuff up, and the output is just out of control. In the end, I still had to manually fix almost everything anyway. Plus, I refuse to share my personal data with big tech companies who just sell it.

So, I decided to build my own platform where I have full control over my data and can do serious research without relying entirely on cloud LLMs: Introlix.

You can see a quick video demo of how it works right in the repo README.

What works right now:

- It’s a clean workspace (looks a bit like Google Docs) paired with a dedicated research engine to keep notes organized.

- Built-in web scrapers and a local database layer so your data stays on your machine.

- Right now it connects to APIs, but I’m actively adding support for local hardware execution over the next couple of days so you can run it 100% offline.

The Future Vision:

When I started this project, my goal was just to build it as a research platform. But now I realize I need a lot of different tools, such as a separate audio processor to remove stutters and clean audio. I haven't started working on this audio feature yet, I'm just making the plan to build it. It won't use an LLM model, it will use a deep learning model instead.

This is where I want to take the platform: I want to shift it from just a research desk into a tool platform. It will host many different ML tools to solve different problems. Users will be able to select exactly what tool they want to use from the UI, and only that specific tool will be downloaded. No extra bloated stuff will touch your disk. You can see the roadmap in the README for full info.

Why I'm sharing this:

Look, I know most of these tools already exist scattered across the web. But they aren't unified into one single platform how I imagine it, and almost none of them keep your data safe.

This is a highly personal project, but I wanted to show it online to get real feedback from people who actually care about self-hosting and privacy. It's 100% open-source and open to contributors. I’m a student and I really want to learn how to manage a project at scale, so if you want to hop in and use it or help build it, you are incredibly welcome.

It’s definitely not perfect, but it’s completely usable and self-hostable today via Docker.

Repo: https://github.com/introlix/introlix-app

Let me know if you have any questions in the comments, or just drop me a DM.

Thanks!

Hello, First of all I'm a student and I use fedora Linux not windows. But I do have to share my pc with my brother and they want windows because they do video editing.

On my computer before I was running windows 11. Windows 11 was running well along with fedora. But 6 months ago I got an update in windows 11 and then it started to crash. And also my brother was not using my pc due to his studies so I told him that I'll uninstall windows right now but will install again when you need it again.

Now, he told me to install it again and so i removed the fedora drive and installed windows on another drive and it ran but when I restarted the pc I got black screen saying something went wrong and will restart my pc and it was stuck in a loop. So, I tried a lot and somehow I got on my computer and updated my pc. Then that error was gone but then windows started lagging and I tried a lot of things updating and someone told me this update causes the problem and when I removed the update I went back to the same boot loop problem.

In the end I installed windows 10 as he needs windows only for video editing. Windows 10 runs well but windows 11 does not run well.

Microsoft has totally broken windows.

Hello, I'm starting college soon and thinking of which degree to do for a robotics software engineer. I'm interested in automation. I want to build a system not just a humonoid robot but any system that is related to automation. And I tried ros2 and gazebo and I did find it interesting and also I find eps and microcontrollers interesting.

So, mostly I want to work on the software heavy side but some won't care if there are some circuits.

I can choose between Computer Engineer or Electronic Systems.

Here is the Electronic Systems syllabus: https://study.iitm.ac.in/es/academics.html#AC1

Now, by seeing this syllabus can anyone answer my question, should I do electronic systems or computer engineering.

Thanks for reading and I hope I'll get my answer.