What If We Could Stop Over 1.2 Million Car Accidents From Happening Each Year?

• Khaled-Alameer Abdelnasser, Computer Science, University of Delaware

• Weisong Shi, CISC, University of Delaware

According to NHTSA, each year in the US alone there are over 1.2 million car accidents caused by adverse weather conditions. This is a major setback in automotive transportation as the weather is out of our control yet many people rely on their car to get to and from work, school, along with many other important day to day activities. This has also been a major weakness holding back autonomous vehicles from being readily available even in today’s age. Why is that? In the sphere of autonomous vehicles there are 5 stages of autonomy that characterize how capable a car is to drive itself. Currently most ‘self-driving’ cars sit at Level 2-3 which means there is a large gap in what’s in the market and where we need to be to provide a safe experience for consumers. Self-driving cars that are being deployed on the road have only been tested in very contained, dry, and favorable climates which simply isn’t realistic for a majority of people. This is where my research comes in as I aim to provide an affordable and easily integrable system that monitors road conditions and adjusts a vehicle’s actions accordingly. Most autonomous vehicles use cameras to read their environment for navigation, speed, and overall decision making performance. Using the captured imaging of these cameras I created a small scale AI model that recognizes when the current weather condition is less than ideal and caps the vehicle’s velocity at a safe driving speed. Using the minimum hydroplaning speed, along with other measured factors, I calculate a maximum speed the vehicle can drive at which takes into account the vehicle’s tire pressure, and other physical features.