seliaste/mindsolver

Alternative software for the Mindcuber-style robots using the ev3dev OS.

Mindsolver

About

This is a software alternative for the Mindcuber-style robots using the ev3dev OS.

It uses statistical classification and mathematical optimization in order to achieve a better rate of success at scanning the cube.

How to run

IMPORTANT: The program currently has hardcoded values for port positions.
Therefore, the required setup is: Sensor arm motor on port B, base motor on port C and flipper motor on port D.
Color sensor can be put on any input port.

Download an executable from the latest release and run it directly on your ev3dev shell.

Alternatively, you can compile Mindsolver yourself

This guide assumes you want to cross-compile from an x86 computer.
Compiling directly on the robot is possible but slow and unsupported

You first need the rust toolchain. This project has been tested for the 1.78.0 stable toolkit

To build mindsolver for a device running ev3dev, simply run:

cargo build --release

The --release flag is optionnal but is recommended for slower compile time but faster execution time

Then copy your file on the robot with scp:

scp target/armv5te-unknown-linux-musleabi/release/mindsolver robot@YOUR_ROBOT_IP:/home/robot/mindsolver

Then, using ssh, simply run the executable using brickrun:

brickrun -r ./mindsolver

Note that on the first run, a cache file will be generated. Scans will be saved in the scans directory if the --save option is used.

Arguments

This software supports some command-line arguments to fine-tune your experience.

Usage: mindsolver [OPTIONS]

Options:
  -f, --file <FILE>            File source if using a previous scan file. Will skip scan
      --iteration <ITERATION>  Number of color sensor scans per facelet [default: 5]
      --movement <MOVEMENT>    Movement between each color sensor scan [default: 8]
      --sleep <SLEEP>          Sleep duration between each color sensor scan (in ms) [default: 20]
  -n, --nosolve                Disables the solution application
  -s, --save                   Enables saving scan to file
  -h, --help                   Print help
  -V, --version                Print version

Run without hardware

You might want to run the program on your own desktop computer without hardware.
In which case, if you use both -f and -n, this program will skip hardware initialization.

For example, cargo run --target x86_64-unknown-linux-gnu -- --file scan_test_files/solvable.txt --nosolve

mindsolver's Issues

Increase fixer success rate

Right now there are 2 different fixing algorithms: the bruteforce and non-bruteforce one. Bruteforce has about 75% success rate with 25% false positive, and non-bruteforce has about 50% with few false positives and often no solutions found.