Learn more about our project
Cycling usership in Canada has increased significantly in recent years. More specifically, in Victoria and the surrounding area bike users have doubled in population since 2015 and the amount of cyclists that use their bike as a regular method of travel has significantly increased. The AR bike helmet was designed and built for the purpose of helping not only the Victorian community but the cycling community as a whole. We feel that one of the best ways to encourage cycling in transportation and recreation is to help riders feel safe on the road. This project was driven by our team's primary focus on reducing bike & car related crashes as a means to promote biking for transportation in urban areas. In recent studies it has been reported that many crashes were due to bike riders navigating in traffic unaware of vehicles approaching them from the rear. With the increasing number of cyclists on the road year to year, evidently there is an ever-increasing amount of accidents on our roadways. Furthermore, the development and integration of E-Bikes in modern society has allowed for a cyclist's maximum speed to greatly increase and as a result a greater emphasis is put on rider safety. Thus, there is a growing need for technological intervention in the safety and experience of an everyday bike rider. With the previous point in mind and in the scope of our project goals it is important to our team that this idea could encourage the development of technologies with similar purpose and application.
Advanced driver-assistance systems (ADASs) have become a salient feature for safety in modern vehicles. They are also a key underlying technology in emerging autonomous vehicles. The technology developed in driver assist cars has helped prevent human error in car crashes by letting drivers know when cars are approaching them. This technology has been proven to work for cars, greatly decreasing the number of crashes and deaths related to human unawareness. Similarly, the ADAS uses ML technologies to increase the awareness of the operator. The transition of this technology to a bike user would evidently have the same effect. The AR bike helmet has been developed with the idea in mind that the technology has already been developed for cars but hasn't been developed for bike users. The helmet would allow cyclists to remain focused on navigating roadways while increasing their spatial awareness with the helmet's blind spot indicator. When a vehicle is approaching the user from the rear, the helmet will notify them via a projected indicator in the rider's field of view. Considering E-bikes have also increased the maximum speed at which a normal cyclist could travel there is an even greater need to develop tools to assist cyclists in regard to enhancing their safety and overall road experience.
Learn more about us
Percent of people in Victoria that use bicycles in a typical week
Daily Average of Bicycle users at UVic south campus entrance
Percent of transportation by single occupancy vehicle in Victoria area
Yearly average of bicycle crashes in British Colombia
Our Design
Our approach to the hardware design for this project can be more easily explained under three categories; computer vision (CV), heads-up display and power supply. The
computer vision aspect of this project is intended to identify and track motor vehicles approaching the cyclist from the rear. The CV module was trained to recognize
the front facing view of a car and identify it as left or right side approaching. This information is then conveyed to the Arduino Nano microcontroller via I2C
communication. The microcontroller is then responsible for signaling the OLED screen to display an arrow with the direction as specified by the HuskyLens. The OLED
screen is positioned so that the image being displayed is reflected off of a tinted lens into the rider’s field of view. Each component in our design is powered by a
single 18650 LiPo cell taken from an existing power bank originally proposed for charging cell phones or other devices via USB. Each hardware component was in part
selected for its ability to be discreetly integrated into the current helmet design. It was important to the designers that such wearable tech provides its service in
the least intrusive way possible.
Click on the boxes below to explore the helmet in greater detail.
What we learned from our project
The outcome of our project revealed a working prototype which was tested in real world scenarios. The helmet successfully detected vehicles approaching from the users rear and effectively notified the rider with a virtual arrow image in the field of view. The advent of augmented reality technologies for cyclists could have a profound effect on cyclist usership, the overall physical health of a population and mitigation of traffic congestion. Throughout the course of this project many challenges and discoveries presented themselves. In this many avenues for improvement and future work in the design were realized. Ultimately, we would like to see that the development of such a project could act as a precursor to future technologies with similar goals and motives within the cycling community. The final prototype for this project was successfully created and tested after the intense design and assembly was accomplished. The code was written and successfully tested with the aid of a serial monitor to ensure the I2C communication bus was functioning between the HuskyLens, OLED display, and microcontroller. The Huskylens was then trained to detect the front of a car and the operation of the microcontroller’s object processing routine was validated. The object's approach was correctly determined and the corresponding indicator symbol was monitored on the OLED display. The functionality of the HUD was then assessed by wearing the helmet and triggering an object detection manually. The HUD was found to function correctly in evening and lower light situations, however, when the ambient lighting levels are too bright, the reflected image is overpowered and lost to the rider's vision. The battery pack’s function was tested by powering up the complete and running system with it and leaving it running until the battery was drained. The single 18650 cell supplied plenty of power storage for testing and prototyping purposes by keeping the device powered up for up to 5-6 hours without recharging. Overall, the device successfully increased riders' awareness on the road by correctly detecting oncoming vehicles and indicating the information to the rider. The safety of the rider was in turn increased due to the added blind spot detection, though, there was not enough time or testing data to justify the claim that the device reduces the amount of collisions between bikers and vehicles. However, a fully functioning wearable, blind spot detection, AR bike helmet prototype was successfully designed, created, and tested by the group.
What We Are Looking to Improve
Display a variety of features to the user such as speed, temperature, distance of approaching vehicle, etc
The electronics can be made more discrete to obtain a more integrated look
Improve the quality of projection and make the projection be percieved as farther away
Develop an ML model suitable to detect not only motor vehicle but other cyclists or pedestrians as well
The indicator can be hard to see in bright outdoor light
Outdoor brightness can be hard on the computer vision hardware (camera)
Acknowledgements
We would like to extend a thank you to our supervisor, Dr. David Capson, for his insight into computer vision and guidance throughout the project.
Thank you to Alex Navarrete for his photography and videography contribution to our presentation and website.
Thank you to Ardeshir Shojaeinasab for his feedback on reports and support.
References
“CRD - Regional Cyclist and Pedestrian Count Program,” Data.eco. [Online]. Available: https://data.eco-counter.com/ParcPublic/?id=4828#.
City of Victoria, "2015 Annual Collision Stattistics Report" City of Victoria Engineering Department, Victoria, 2015.
“Bike counts,” CRD, 07-Nov-2013. [Online]. Available: https://www.crd.bc.ca/about/data/bike-counts.
Free data visualization software. [Online]. Available: https://public.tableau.com/app/profile/icbc/viz/QuickStatistics-Crashesinvolving/CrashesInvolving.
“Yoobao® Thunder 13,000mah universal dual-USB battery pack with Transformers symbol, led capacity indicators&flashlight for iPhone/Android phone(Samsung, Google, LG..)/blackberry/nokia/etc., tablet, iPad/ipod, MP3, PSP and more - black,” Amazon.ca: Electronics. [Online]. Available: https://www.amazon.ca/Universal-Transformers-Indicators-Flashlight-BlackBerry/dp/B00G6C34UE.
“HuskyLens – An AI Camera: Click, Learn, and Play | Vision Sensor | DFRobot Electronics,” www.dfrobot.com. https://www.dfrobot.com/product-1922.html
“IZOKEE 0.96" I2C IIC 12864 128X64 Pixel OLED LCD Display Shield Board Module SSD1306 Chip 4 Pin for Arduino Display Raspberry Pi 51 Msp420 Stim32 SCR (Pack of 3pcs, Yellow-Blue-IIC) : Amazon.ca: Electronics,” www.amazon.ca. https://www.amazon.ca/gp/product/B076PNP2VD/ref=ppx_yo_dt_b_asin_title_o01_s00?ie=UTF8&psc=1
“Reflection and Refraction,” Let’s Talk Science. https://letstalkscience.ca/educational-resources/backgrounders/reflection-and-refraction
Contact Us
Victoria BC, Canada