Presentation Title

Self-Driving Remote Controlled Car

Faculty Mentor

Meng-Lai Yin

Start Date

18-11-2017 10:00 AM

End Date

18-11-2017 11:00 AM

Location

BSC-Ursa Minor 85

Session

Poster 1

Type of Presentation

Poster

Subject Area

engineering_computer_science

Abstract

In this project, a self-driving remote controlled (RC) car is built, which can drive autonomously by implementing machine learning functions through a neural network model. The car is able to navigate itself along a sidewalk while maintaining its course. Capabilities such as collision avoidance and real-time steering are provided. This project successfully demonstrates the integration of machine learning and microcomputer engineering.

The RC car is equipped with a microcomputer (e.g., Raspberry Pi) and a wide-angle camera. The Raspberry Pi processes the images obtained from the camera and predicts the steering angle. The Raspberry Pi also controls a servo so that real-time steering capability is achieved.

The essential machine learning component in this project is the implementation of the convolutional nine-layer neural network model. A dataset of 3,000 images with corresponding steering angles is applied to train the model using the open source software library Tensorflow.

Keywords: machine learning, neural network, collision avoidance, microcomputer, dataset, servo, Tensorflow

This document is currently not available here.

Share

COinS
 
Nov 18th, 10:00 AM Nov 18th, 11:00 AM

Self-Driving Remote Controlled Car

BSC-Ursa Minor 85

In this project, a self-driving remote controlled (RC) car is built, which can drive autonomously by implementing machine learning functions through a neural network model. The car is able to navigate itself along a sidewalk while maintaining its course. Capabilities such as collision avoidance and real-time steering are provided. This project successfully demonstrates the integration of machine learning and microcomputer engineering.

The RC car is equipped with a microcomputer (e.g., Raspberry Pi) and a wide-angle camera. The Raspberry Pi processes the images obtained from the camera and predicts the steering angle. The Raspberry Pi also controls a servo so that real-time steering capability is achieved.

The essential machine learning component in this project is the implementation of the convolutional nine-layer neural network model. A dataset of 3,000 images with corresponding steering angles is applied to train the model using the open source software library Tensorflow.

Keywords: machine learning, neural network, collision avoidance, microcomputer, dataset, servo, Tensorflow