Universita' della Svizzera Italiana - Robotics Final project

Real robot place recognition using Convolutional Neural Network (CNN) and ROS

Authors: Andrea Bennati & Alessio Della Libera

This repository contains our final project for Robotic Course.

The goal of this project is to use the Thymio robot to predict in which place it is.

Python Version: 2.7

The libray used are:

• pytorch v1.1
• openCV (cv2)
• matplotlib
• numpy
• ROS (molodic version)
• keyboard library

The following code has been tested only on Ubuntu 18.04.

Control Usage

python main.py main.py [-h] [--action {random_walking,human,stop}] --name NAME [--save_image SAVE_IMAGE] [--path_save_image PATH_SAVE_IMAGE] [--predict PREDICT] [--render RENDER] [--record RECORD] [--path_model PATH_MODEL]

Examples

Random Walking (moving and save images)

python main.py main.py --action random_walking --name thymio23 --save_image True --path_save_image /path/to/directory

Human Control (moving and predict)

python main.py main.py --action human --name thymio23 --predict True --path_model /path/to/saved/model

CNN Usage

python Thymio_cnn.py [-h] --src SRC [--path_model PATH_MODEL] [--action {train,display}]

Examples

Train the CNN

python Thymio_cnn.py --src /path/to/images/ --action train

Display Images

python Thymio_cnn.py --path_model /path/to/saved/model