Home

Welcome to the AutoRCCar wiki! Working model of Self-driving car using Convolutional Neural Network, Raspberry Pi and Arduino Aditya Kumar Jain Electronics and Communication Department Dharmsinh Desai University Gujarat, India [email protected] Abstract— The evolution of Artificial Intelligence has served as the catalyst in the field of technology. We can now develop things which was once just an imagination. One of such creation is the birth of self-driving car. Days have come where one can do their work or even sleep in the car and without even touching the steering wheel, accelerator you will still be able to reach your target destination safely. This paper proposes a working model of self-driving car which is capable of driving from one location to the other or to say on different types of tracks such as curved tracks, straight tracks and straight followed by curved tracks. A camera module is mounted over the top of the car along with Raspberry Pi sends the images from real world to the Convolutional Neural Network which then predicts one of the following directions .i.e. right, left, forward or stop which is then followed by sending a signal from the Arduino to the controller of the remote controlled car and as a result of it the car moves in the desired direction without any human intervention. Keywords— Artificial Intelligence, Self-driving car, Convolutional Neural Network, Raspberry PI, Arduino I. INTRODUCTION Globally speaking, nearly 1.3 million people die in road crashes each year, on average 3,287 deaths a day [1].And talking about India the number of people who were killed in a road accidents in 2013 alone were 1, 37,000 [2].Speeding, talking over phone, drunk driving and breaking traffic rules are the root causes behind these accidents and the statistics are rising day by day which is now becoming a major concern. No matter how hard we try to create awareness regarding traffic rules and safety that has to be followed while driving, accidents are still occurring and aren’t showing a sign to stop. Though human errors can never be eliminated, but accidents can definitely be stopped. And in this case technology has surely come to our rescue. Starting from the very early radarbased collision detection to present day’s technology, the advancement and improvement in this technology had seen an exponential growth in recent years. Self-driving cars is the one of the most discussed technology of current scenario. What was once imagined is a reality now. The definition Self-driving cars is a car which promises to take the traveler to their destination with minimal human control while taking safety as its first priority. Many companies throughout the world are making a serious and continuous efforts to make driving a safe and risk free process and have started building prototypes for the same. Amongst these companies are Google, Tesla Mercedes and many more who have built a successful and functioning prototype and are planning to release a model in the upcoming years. Self-driving cars are expected to have faster reflexes than humans, make more reliable judgments, thus avoiding mere faults which causes accidents at the first place. Apart from saving precious lives, other advantages these technology gives is better traffic flow regulation because unlike humans these cars ride with proper traffic rules, making rides smooth and congestion free. Self-driving cars can also help in tackling parking space issues by allowing to create a taxi/pooling service for the unused cars and by unused car we mean to say the car that is either staying for few hours while the owner is at work or the car that is in the garage while the owner is out for a vacation. Thus we could make better use of land instead of using it for parking space. The basic model of any Autopilot system involves radar, a front-facing camera, a digitally-controlled digital braking system, and long range ultrasonic sensors located around the car. Radars enables the detection of vehicles and other moving objects around the car, front facing camera helps to detect and recognize objects like cars, trees, driving lane, humans, traffic signals and other important data. All these information are taken in real-time environment and are fused into a learning network which then predicts the car’s response accordingly.