Install UV (Package/venv Manager)

MacOS/Linux

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

PypI

pip install uv

Note position_nn is the only directory setup with uv.

Linting

uvx run ruff

Running Scripts

cd position_nn
uv run process_data.py
uv run train.py
uv run predict.py <args>

For example:

uv run predict.py simulation.csv 0 0 0 0 0 0

WIP: args, functionality

Directory Structure

position_nn/: Position model. Given positions/rotational data and potentiometer inputs over the same timeframe, output position predictions with a simple LSTM neural network.

• raw_data/: CSV naming format ###{ID}-MMDD{TEST_DATE}-####{START_FRAME}-####{END_FRAME}-###{FPS}
• processed_data/: Post-processed dataset (by process_data.py)
• process_data.py: WIP
• predict.py: WIP
• train.py: WIP

Legacy Folders

acceleration_nn/: Acceleration model. Given positions/rotational data and fixed potentiometer value as input, output acceleration with a simple MLP model. old-temp/: Kinematics (steady-state) model. Given positions/rotational data and fixed potentiometer value as input, output average acceleration with kinematic equations.

TODOs:

WIP (heh)

LSTM Neural Network Quick Overview

WIP