mjwhite8119 / romiroadfollow

Romi Road Follow

This project implements road following on a Romi robot using a deep learning neural network. There are three main components for this project:

• Data Collection - collects images and steering values on the Romi for training. See the DataCollection README.

• Training - Uses the images and steering values to train the neural network model. See the Training README.

• Deployment - deploys the trained model on the Romi to send data back to the Java WPI program and control the robot. See the Deployment README.

The WPI Java program can be obtained from from the RomiExamples repository on Github. The example program used is called BasicML

Install Tensorflow on M1 Mac

conda create --name tf-env python=3.8
conda activate tf-env
conda install -c apple tensorflow-deps
pip install tensorflow-macos
pip install tensorflow-metal
conda install -c conda-forge -y pandas jupyter

python3 -m pip install -r requirements.txt

Tensorboard needs a lower version of markdown, therefore:

pip uninstall markdown
pip install markdown==3.1.1

LabelImg Install

Using Conda. Create a virtual environment in conda and activate it:

conda create -n venv
conda activate venv

Install pyqt using conda:

conda install pyqt

Install lxml using pip:

pip install lxml

Clone labelImg:

git clone https://github.com/tzutalin/labelImg.git   
cd labelImg
make qt5py3

Run LabelImg:

python labelImg.py

Referencies

• Tensorflow Basics

• Frozen graph format how-to

• Depthai OpenCV AI Kit