Bike Project

Our goal is to help make biking safer in our community.

Embedded Systems Design / 18549 / Carnegie Mellon University

Jeffrey He

Biking, every day, all season.

Marc-Daniel Julien

Literally running around town.

Vishnu Razdan

A chill dude who likes to chill and make bikes safer.

Francisco Rojo

The trophie white male.

Concept

This project will allow Bike Pittsburgh to gather data on the proximity of cars to cyclists on the road. This information will help plan bike lanes and routes.

The prototype will be a sensor package that can be attached to any bike. While originally planned to use multiple ultrasonic sensors, instead we now use a single Lidar system in combination with Bluetooth communication and a mobile phone. The end result is that the gathered data will be stored on a web server to be analyzed by interested parties.

Motivation

As of today, there is no widespread, easy to use product that enables city planners to quantitatively measure the safety of bikers throughout a city. Some streets are safer than others for cyclists, and we want to know where. This data is valuable for the city because it can be used to help improve the quality of biking throughout the city. Safer bike lanes, additional bike lanes, and alternate biking routes can be suggested with this data.

Using our prototype system, cities will be able to collect data of how close a vehicle is to a biker at a specific GPS location, and use this data to make biking safer.

Competition

Similar existing products are:

Vanhawks Valour Smart Bike

  • This product is a connect smart bike that provides:
  • GPS navigation, including turn by turn instructions via LED lights on the handlebar, with data fed by your smart phone via LE bluetooth communication.
  • Haptic feedback of incoming traffic or other hazards in your blindspot
  • A rich set of data, including time, speed, best time, distance, and calories
  • How our product differentiates itself:
  • Our product provides much of the same functionality but is not tied down to a specific bike - it is bike independent so users can still just use their own bike
  • Our product is also much cheaper, since they only have to purchase the module instead of an entire bike

BikeShield App

  • This product is an app that alerts drivers when bikes are nearby so you can be more aware and more cautious. The app on the biker’s side alerts the app on the car or bus driver’s side when they are in proximity
  • Their product depends on every car or bus in proximity to also have the app installed and running, which is quite unrealistic. Ours is independent of other people’s choices.

Butch Smart Bike

  • This product is a 55lb bike at $2000 - $4000 dollars that includes forward radar collision detection, rear view CV hazard detection, and is electric powered.
  • Our product is not 55lbs heavy, is not over expensive, and is bike independent.

Requirements

Primary

PR-1 The System shall be able to identify all cars that come within 4 meters of the rider.
PR-2 The System shall identify cars within a +/- 0.25 meter accuracy.
PR-3 The System shall not make any false positive identification of cars.
PR-4 The System shall identify each identified car’s GPS location to within +/- 50 meters.
PR-5 The System shall work in the all weather and lighting conditions.
PR-6 The System should be able to operate for at least 60 minutes before being recharged.
PR-7 The data acquired should be easily viewable on the web.

Secondary

SR-2 The System shall alert the rider of an oncoming car in their blind spot within 3 seconds.
SR-2 The System shall alert the rider of dangerous roads.

Technical Specifications

Hardware

      
Part NameQuantityUnit PriceTotal PriceDescriptionSourceLink
HRXL-MaxSonar-WR4$99.95$399.80Ultrasonic Range FinderSparkfunhttps://www.sparkfun.com/products/11724
CC25401$4.61$4.61SOC with BLETIhttps://store.ti.com/CC2540F256RHAR.aspx
ATSAMB111$45.00$45.00SOC with BLEAtmelhttp://www.digikey.com/product-detail/en/atmel/ATSAMB11-MR210CA/ATSAMB11-MR210CA-ND/5638846
ADXL3621$14.95$14.95Triple Axis AccelerometerSparkfunhttps://www.sparkfun.com/products/11446
PCB1~$75~$75Custom PCB  
AmazonBasics Portable Power Bank1$19.99$19.99External 5600 mAh power supplyAmazonhttp://www.amazon.com/dp/B00LRK8IV0/ref=twister_B01AI55GD2?_encoding=UTF8&psc=1
Miscellaneous1~$50~$50Resistors, capacitors, clock oscillator  
TOTAL:  $624.30  

Software


Embedded Code C and assembly.
Website Python (Django web framework) and JavaScript
Android App Java

Protocols


Bluetooth Communication from sensor package to phone.
Cellular data Communication from phone to web server.

Updates

March 16th: We got our first data using the ultrasonic sensors! Sitting on the sidewalk on Forbes we managed to record cars passing by, with the sensors plugged into an Arduino as our design is not yet finalized. The decrease in the sampled voltage indicates how close the vehicles got to us.
March 28th: We ran into our first major issue. The ultrasonic sensors are too unreliable to be used to detect cars. We talked to Professor Rowe who bought us this very nice Lidar system to use instead! We have just started to learn how to use the system, but things are looking promising.
On the left is Marc and Jeffrey on April 18th getting some street data using the Lidar system. At this point we still do not have our PCB as we recently learned that the toolchain for our original TI chip costs about $750, which is $50 over our total budget, so instead we are using Marc's laptop until we get the Atmel chip.
On April 21st we selected the lucky bike to be used to have our package mounted upon for our complete integration, shown on the right.
April 25th and 26th: We finally have our complete PCB! After getting Aaron to help flow the chip onto the board, we finished soldering the rest of the components onto the PCB at Jeffrey's house. Now we begin the mad dash towards our demo for complete system integration.
We all met at Jeffrey's apartment later that night and put together our complete system. However, tragedy struck when we burnt our PCB accidentally. Since this is the final stretch, we decided to use a Raspberry Pi 3 as it has built in Bluetooth. The row above shows us working in Jeffrey's house, and on the left is us doing street testing.
We continued to run into problem after problem, as now using out setup on the left almost every single time a car would pass our system on the bike, the Lidar sensor and Raspberry Pi would crash! We later realized this was because of a brown-out, but at the time we had no clue why this was happening. In order to make forward progress, we swapped the Raspberry Pi for Jeffrey's laptop, as shown on the right.
By Demo Day we had a working system that would detect cars and note how close they got to you while biking, with the caveat of still needing the laptop as we only had our burnt PCB and irregularly failing Raspberry Pi. We presented and spread the word about bike safety!