Progress Devlog 2 - Phoenix 1:

So basically, YOLO didn’t really work out, so I had to delete it because I didn’t have enough RAM on my Raspberry Pi. Instead, I got ROS 2(Specifically ROS 2 Jazzy, an older version released for Ubuntu 24.04.02). For my original version of main.py, which is a system that launches 4 separate files(Ultrasonics and Motors, both arduino and python). Now using ROS 2 I built a separate mapping system called Rviz2, which is a very versatile mapping system that can spawn things like walls and stuff, and its pretty cool. But then I had to create a bunch of new files, like “phoenix_config.rviz” and I needed to create a new launching system called ./launch.sh, which booted the necessary rviz terminal commands and my main.py simultaneously. I also made something called an RF Model, also known as Random Forest.

What’s Random Forest?
It’s basically a module that you can import via python that uses ultrasonic readings and turns them into basic YOLO detection: Like “Corner” or “Left Wall” and that RF communicates to my Rviz2 display, telling it to snap walls or stuff like that when it detects “Corner” or “T-Junction”.

For my RF Model, I used a CSV file which uses my binary encodings(1 for blocked ultrasonic and 0 for not blocked ultrasonic). It was pretty simple, and with my 4 ultrasonic sensors I was able to create 16 total types of obstacles through my RF:

0,0,0,0,Path Clear
1,1,0,0,Left Corner Trap
1,0,1,0,Right Corner Trap
1,1,1,0,Dead End
1,1,1,1,Total Box Trap
0,1,1,0,Narrow Choke Point
0,0,0,1,Back Blocked
0,1,0,0,Left Wall
0,0,1,0,Right Wall
1,0,0,0,Front Wall
1,1,0,1,T-Junction Right Open
1,0,1,1,T-Junction Left Open
1,0,0,1,Fork
0,1,1,1,U-Tunnel Forward Open
0,1,0,1,Left Lane Guard
0,0,1,1,Right Lane Guard

I also created a few JSON files with timestamps and stuff to log where the walls were created in Rviz2 and when and stuff like that:

[
   {    
        "label": "Front wall",    
        "timestamp": "Jul 16 15:49",    
        "coordinates":   [      
            3.0,      
            0.0    
        ]  
    },
    {    
        "label": "Front wall",    
        "timestamp": "Jul 16 15:49",    
        "coordinates":   [      
            3.0,      
            -0.25    
        ]  
    }
]

So that’s basically what I spent my 9 hours of precious time on. It was Lwk such a pain but it was worth it.