Modero: Building Better Intersections with AI and Electromagnetism
Problems with existing infrastructure
In the United States alone, there are more than 2.5 million intersection accidents annually. This accounts for 40% of all car crashes. Even though there are many road safety laws and technologies designed to protect us, there are still two people who die in the U.S. everyday from people who do not stop at red lights. With the computer vision and AI fields increasing so rapidly and overlapping with so many other sectors, it is baffling that there are no scalable solutions to this problem yet. Cities are starting to roll out smart traffic lights to decrease traffic congestion. The issue with this is that it only solves part of the problem. A smart traffic light uses AI and an array of sensors to route vehicle and pedestrian traffic. The problem is, even if a smart traffic light detects a crash that is about to occur, it has no way of helping to avoid the crash. Modero aims to reduce the amount of casualties due to intersection crashes.
The effects of intersection accidents extend far beyond the casualties that occur. In addition to requiring road maintenance, intersection accidents usually slow or halt the flow of traffic and as a result of many cars idling, carbon emissions increase. According to NewScientist, CO2 emissions are projected to reduce by as much as 6.5% when smart traffic lights are fully implemented. Using our solution, we expect this percentage to rise even higher.
How it works
We will be modifying the existing smart traffic light infrastructure with our AI collision detector and bluetooth transmitters. Our brakes will be in constant communication with our AI-powered traffic lights. Each set of brakes will have a bluetooth transmitter and receiver which will be used to communicate with the traffic light and other cars. Our approach to connectivity is similar to Apple’s Find My iPhone. When an iPhone is reported lost, it pings all other Apple devices in the area to receive a connection to the internet. Similarly, cars that are too far from the traffic light will get an indirect connection via other cars using the brake’s bluetooth receiver. The nature of this network is that the more cars that use Modero brakes, the stronger the system becomes. Modero means “slowing down” in Latin. When one of our AI traffic lights identifies a collision that is about to occur, all the cars on every side of the intersection will be slowed and then stopped until it is safe to continue driving. Factors that the AI will be looking for include the speed of oncoming vehicles, the amount of traffic, pedestrians, motorcyclists, nearby vehicles, and potential obstacles, such as ice, snow, or other debris. All of these factors will be weighed by the AI and the intersection will be slowed proportionally. The brake will store information about the car’s weight and the traffic light will be able to send the stopping signal so that all the cars stop at the same time.
After reading this far, you may be thinking: What about motorcycles? In the case of an impending motorcycle accident, our system would be able to stop all cars in the vicinity and minimize the force of the impact.
Feasibility
Computer vision and AI have become very advanced in the past couple of years. With Tesla and Waymo having major breakthroughs in the self driving car industry, where classification of other cars and identification of threats is critical. This can be seen as a proof of concept for our threat detection and prevention as it is a much simpler task than controlling the steering of the car. We plan to use the deepSORT and YOLO3 algorithms to track the vehicles’ speed and trajectory. We will also be using a classification (pattern recognition) algorithm to perform image filtering as well to identify and accommodate emergency vehicles. The AI component is undoubtedly achievable due to the sheer amount of research that has already been conducted in this field.
That being said, there are a couple things that need to change in order for our product to work. First of all, we need to have stronger and more affordable electromagnets. There is currently no known limit to the strength of an electromagnetic field. According to our calculations, to be able to stop a 2000kg car (much heavier than the average car) within 10 meters that is moving at a speed of 90km/h (much higher than speed limits for intersections), we will need to have four electromagnets with 15,625 N of force each (totaling 62,500N for the entire car). Using traditional brakes, the average car needs 65m to stop completely moving at a speed of 90km/h. This is the reason that we are choosing to use electromagnetic brakes in our system. Another issue that needs to be overcome is the form factor of the electromagnets. Right now, the electromagnets will add a lot of bulk to the look of the car and it may take a while for the technology to be compact enough that it can sleekly blend into the shape of the car. In addition to this, because we are piggybacking off of existing smart traffic light systems, these systems will need to have a higher adoption rate nationally. Currently, only 3% of America’s traffic signals are “smart traffic lights”. Although this seems small, there is a growing amount of cities that have plans to install smart traffic lights. The number of smart traffic lights has grown from 4,500 in 2009 to 6,500 in 2014. Fortunately for us, the largest concentration of smart traffic lights happens to be in Los Angeles.
Location & Timeline
We have chosen to begin the launch of Modero in Los Angeles, California. This is because LA has the highest rate of injury-producing and fatal traffic accidents in the United States. In Los Angeles, at least one person is killed in a traffic collision every 40 hours. LA has around 150 crashes per day. 40% of car crashes involve an intersection incident. This equates to 60 intersection related incidents per day. At least 93% of vehicle collisions are due to driver error. As you can see, there is a real need for our product. If we can remove human error from the intersections of LA we will save countless lives.
Cost Estimation
Traffic accidents cost California $4.48 billion per year. Our project’s total cost equals to $45.3 billion. A traditional brake set would cost the consumer $300. If we charge the consumer the same price for Modero brakes, the total price for the program becomes $40.8 billion. The unit cost of our brake package will be $3020 for the government, $2720 if we charge consumers a $300 fee. This breaks down to $2000 for the electromagnets, $1000 of labor cost/profit, and $20 for the bluetooth transmitter/receiver. Electromagnets are pretty expensive right now. To get a magnet with the max pull that we need (4000kg), it costs $500 per wheel. Bluetooth transmitters/receivers are much more affordable. They cost around $20 for 400 meters of range. It will take the government roughly 10 years to recover the $40.8 billion spent on this program ($4.48 billion saved annually).
Metric for Success
This project will be deemed successful if we are able to save at least one of the two people who die everyday due to red light runners. This will equate to almost 400 people per year! Financially, if the government can recoup the cost of installing our systems within 10 years, the product will be an absolute success. As a bonus, traffic congestion is projected to reduce and CO2 emissions will see at least a 6.5% decrease.
Demo
If you are interested in seeing a mockup of our braking system check out this video:
