Smart Helmet
Our smart helmet is designed for cyclists that provides useful features and extra safety support. It employs cutting-edge navigation technology and state-of-the-art neural network object detection to steer the cyclists away from danger.
Motivation
● More and more people enjoy cycling because it is healthy and environmentally friendly;
● Bikes generally don’t offer mounting space for assistance devices or smartphone;
● Cyclists are interested in tracking their health status when considering cycling as a sport;
● Few bikes have rear view mirrors, and cyclists are unaware of the traffic behind unless purposely looking back, which imposes danger.
System
Architecture
The whole system mainly consists of three parts:
● The raspberry pi placed on top of the helmet provides voice interaction and navigation; the Google vision kit placed on the rear is incorporated to detect cars and provide safety warnings; a solar panel on top of the helmet provides the power.
● AWS server and database are running on the cloud to process and store client trip information and health statistics.
● All of the user formations can be visualized on our website: http://guosl.com/iot/.
Technical Components
● Raspberry Pi is used as the main controller;
● User portal website presents trip information and health statistics;
● Google API is used to provide audio navigation;
● User data is stored in a cloud database;
● Google vision kit is used to monitor the rear view;
● All components are powered by solar energy.
Prototype
We have developed a prototype for our smart helmet project as shown below. It provides most of the feature in our initial design. We have also made a custom case using 3D printing for the rearview camera and mounted it on the helmet. A solar power bank is placed on top of the helmet to provide power to both the raspberry pi and the rearview camera.
Results
Direction webpage and action buttons to send audio navigation to helmet or finish trip
The audio response from the smart helmet
Rearview vehicle detection (warnings are triggered if the size of the bounding box exceed certain threshold)
References
● AIY Vision Kit from Google. (n.d.). Retrieved from https://aiyprojects.withgoogle.com/vision/
● G. (2018, August 03). Google/aiyprojects-raspbian. Retrieved from https://github.com/google/aiyprojects-raspbian
● T. (2018, October 03). Tensorflow/models. Retrieved from https://github.com/tensorflow/models/tree/master/research/object_detection
● Google Map Platform. Retrieved from https://developers.google.com/maps/documentation/directions/intro
Our Team
Guanxuan Li
M.S. candidate in Electrical Engineering | Concentration: Networks, Communications, Embedded Systems, Internet of Things.
My LinkedIn
Shanglin Guo
MS student at Columbia University with extensive experience in embedded programming and hardware-software integrations.
My LinkedIn
Zhansen Shen
Electrical Engineer | Concentration: Machine learning and Internet of Things.
My LinkedIn
Contact
Guanxuan Li: gl2619@columbia.edu
Shanglin Guo: sg3640@columbia.edu
Zhansen Shen: zs2390@columbia.edu
Columbia University Department of Electrical Engineering
Instructor: Professor Xiaofan (Fred) Jiang