Car-N-sics Blackbox
Website for car-n-sics project for intel hackathon
Use Arduino 101’s 6-axis accelerometer/gyro + machine learning to figure out road conditions and driving patterns
Hardware
- Arduino 101
- SD Card reader/writer
- RealTimeClock
- LCD display
- EM506 GPS Receiver Module
- Arduino Uno (used as GPS proxy)
- Button (used in training mode/data export)
- Beeper (optional)
Installation
cd ml
virtualenv --system-site-packages -p python3 .pyenv
source ./pyenv/bin/activate
pip install -r requirements.txt
Preparing data
cd rnn
. ./pyenv/bin/activate
mkdir -p processed/{test/train}
python prepare.py -i ../raw_data/new_format/17_6_30.CSV -i ../raw_data/new_format/17_7_1.CSV -i ../raw_data/new_format/17_7_2.CSV -i ../raw_data/new_format/17_7_3.CSV -o processed -c 200 -u 2 -l 100 -r 0.6
Training a model
cd rnn
. ./pyenv/bin/activate
python train_rnn.py processed model_666.ckpf
Classifying a csv
cd rnn
. ./pyenv/bin/activate
python classify_em.py -i ../raw_data/new_format/17_7_3.CSV -c 200 -u 2 -l 100 -m model_666.ckpf
Running tests
cd ml
. ./pyenv/bin/activate
pytest -vv
Prior arts
DTW+kNN approach:
http://nbviewer.jupyter.org/github/markdregan/K-Nearest-Neighbors-with-Dynamic-Time-Warping/blob/master/K_Nearest_Neighbor_Dynamic_Time_Warping.ipynb
RNN approach:
https://github.com/guillaume-chevalier/LSTM-Human-Activity-Recognition/blob/master/LSTM.ipynb
by:
