Thanks for your interest in ADetector

The project is called “Real-Time Accident Detection on Michigan Highways”, The team proposes to develop a computerized accident detection system for Michigan highways. The traffic data captured by thousands of video cameras across Michigan can be processed into image pixels using Image Processing Software (IPS). Computer models, including traffic simulation models, simulate different scenarios of road congestion and decrease in road access due to an accident.


Team members are (left to right in photo) Ibraheem Nadeem (Pre-Engineering and Honor student), Maria Hanna (Pre-Engineering and Honor student), Isauro Sanchez (Pre-Engineering student), Omar Dahbali (Pre-Engineering student and President of Engineering Club), and Dr. Hassan Nameghi

The team built the first prototype to prove the concept. Here is a video of the machine. Rolling marbles represent the cars on the road. To learn more about how the machine was made, read here.

Title: Real-time Accident Detection on Michigan Highways

The Problem: Traffic congestion brings about a huge cost for trucking industry in US: over $9.2 billion dollars and a total of 141 million hours of lost productivity in 2013 (, 2014). In particular, vehicular accidents contribute to a large portion of road congestions in Michigan. “Accidents cause traffic congestion and delay to increase and even cause casualties and property losses”(Shi, 2012). Currently, there is no practical solution to detect accident in a real-time. The only reliable solution is that someone should call 911 after an accident occurs. However, according to the bystander effect, the more people present at the moment, the less likely they are to help a person in need. In addition, if someone calls 911, it will be hard to communicate the location. It could get confusing if multiple people call and provide inaccurate information about the location. An Increase in population and car ownership lead to more traffic congestion, disrupted flow, and accident likelihood. Therefore, if the current problem is not alleviated, it may contribute to more social anxiety, traffic delay, loss of lives, fuel consumption, and air pollution. On the other hand,
a real-time detection of an accident and immediate dispatch of police, ambulance, and fire department could save the lives of people and reduce traffic congestion. Recent investments in infrastructure of Intelligent Transportation Systems accommodate collection of big traffic data
for computerized detection of accident.

The Solution: Our solution is to develop a computerized accident detection system for Michigan Highways. The big traffic data captured by thousands of video cameras across Michigan can be processed into image pixels using Image Processing Software. Advancement in computer science and traffic engineering accommodates the technology to estimate traffic speed, traffic flow, and traffic density from image pixels. Traffic density is the number of vehicles in one mile. When an accident occurs, the road capacity reduces. Therefore, traffic density increases while traffic flow and speed decreases. By mimicking the change in traffic characteristics, Torfehnejad & Adamnejad (2014) proposed a simple method for detection of abnormal traffic flow in one freeway segment. However, this method is insufficient for detection of accident in a road network. This is because abnormality in traffic flow could be due to a merge-in arrival traffic or merge-out backup. Previous literature used the analogy that the road network is similar to electric circuits. In this study, it was assumed that road congestion is similar to a resistor, speed is similar to voltage, and flow is similar to current. Advancement in electrical simulation allows quick solution to complex dynamics electric system. Thus, the value of resistors can be recalculated and compared with the actual one for every road. If the actual resistor is much higher than the calculated one, this could indicate presence of an accident. To make the solution a reality, it is required to create an algorithm to differentiate normal from abnormal traffic.

Impacts and Benefits
If the computerized accident detection system is successfully implemented, it will have countless benefits in areas such as economics, society, industry, and science. It will reduce traffic delay, disruption, and casualties after a vehicular accident. In addition, it will contribute to faster movements of goods and people – which directly lead to a better economy. It will benefit society by having a safer and less stressful environment. A reduction in accident response time will also result in less traffic jams, less air pollution and less gas consumption. According to Cox (2000), air pollution increases as automobile speeds decrease; moreover, higher air pollutions are associated with higher traffic densities. Less traffic jams will provide individuals with extra time to achieve more tasks. Science will greatly benefit from implementing our solution because the concept of associating electric circuits to simulate highway traffic for accident detection is an innovative idea. To the best of our knowledge, the use of big traffic data for identifying vehicular accidents on road networks received less attention by prior studies. The Impact of our solution can be measured using various tools such as computer models, survey analysis, and traffic simulation. Using survey analysis, a comparison will be made between previous emergency response times and current ones. Computer models such as traffic simulation models can be used to simulate different scenarios of road congestion and decrease in road capacities due to an accident.