Our Team members are motorcycle enthusiasts and care very much about safety when riding a motorcycle. Even with following all the rules when riding a motorcycle the rider is still at risk of being involved in an accident even if they are not at fault. Usually that is caused by drivers who were not able to spot the rider and end up hitting them.
Our idea came from looking at youtube videos involving motorcycle riders being hit from the back and also being put into this situation in real life pretty often because the other vehicle driver was too fast or did not see them in time.
How it works
We used opensource Computer vision libraries OpenCV and built the system on an android device since being mobile friendly was the goal as well as feasibility. The app spots vehicles behind the motorcycle rider and calculates the speed of the vehicle coming behind them while the rider is at a full stop. If the vehicle behind the rider is too fast and the stopping distance is greater during which the rider is at the risk of being hit, using the pebble watch we alert the user to get out of the way since being on a motorcycle gives the rider a lot more flexibility.
Challenges I ran into
For this project we could've used a radar gun to find the speed of the vehicle behind the rider but the goal was to make it mobile friendly and gives the riders the confidence by giving them control of the situation they might go through. Doing computer vising on mobile platform is also another challenge since mobile devices dont have enough computing power.
Accomplishments that I'm proud of
Calculating speed of an object using computer vision was the hardest challenge for this project as the speed was not always accurate.
What I learned
We learned a few new things about computer vision, Object detecting algorithms and how accuracy is affected, also had a basic refresh of basic physics for calculating velocity of an object.
What's next for Beyekster
Our goals for Beyekster are:
- Make it more accurate when it comes to calculating speed of an object in the camera.
- Use better algorithms to spot vehicles as well as identify them so we can check the weight of the vehicle with a database and have more accurate stopping distance.
- Add new features such as automatic emergency request in case the rider was hit and got hurt.