Flip the breaks!

Flip the breaks!

A system set up for flattening speed bumps for ambulances, to reduce patient risk and save time, using computer vision.

Flip the breaks!

Flip the breaks!

A system set up for flattening speed bumps for ambulances, to reduce patient risk and save time, using computer vision.

The problem Flip the breaks! solves

In this fast world for safety purposes and to prevent accidents taking place on the road, there is a concrete speed breaker placed on roads to limit the speed of the vehicles. But at the same time, they are major obstacles for emergency vehicles like ambulances. The sudden bump or jerk will be harmful to the patient inside. For example, in the case of a pregnant woman or a heart patient. Also, the patients who meet with an accident or any sudden medical issues are taken to hospitals using an ambulance. The speed breaker reduces the speed of every vehicle to a certain range, which causes a time delay.

So, taking a patient’s safety & life into consideration, the speed breakers are an obstacle for safe path to hospital, but for the concern of civilian traffic, the speed breakers shouldn’t be entirely removed.

Challenges we ran into

The main challenges we ran over was the unavailability of proper dataset for this particular task as the dataset contained highly random types of ambulances. This was solved by manual annotation of 600+ images with MakeSenseAi which lead to much better accuracy.
The second challenge we came over during our brainstorming seesion was related to the practical application of flipping the speed breakers. Due to presence of traffic and other vehicles, the timing of the flip is crucial. That is why we have included a traffic density indicator in our system. Only below a certain threshold of traffic density, the flip will come into action. This ensures an extra level of safety in the roads.
Also, in our dataset "ImageNetObject Localized", we found images of flipped ambulances,helicopters and jeeps. These were errors we had to overcome manually with careful selection during our Data Analysis.

Discussion