I NTRODUCTION
This project report aims to document the process involved inbuilding a self-driving car using inputs form ultra-sonic sensor,
IR sensors, and a camera. It involves setting up the devices, es-tablishing communication between them, implementing image
processing using OpenCV4, training a cascade classifier, de-signing dynamic turn signal circuits which would control the
motors or wheels of the car. The intent of the project is that theself driving car can interpret and understand its surroundings
and navigate safely such as it can detect traffic lights, detect different signals like stop signs, detect vehicle obstacles and
perform road lane changes within lane boundaries thus make appropriate driving decisions.
II TEAM MEMBERS
Shivam Kar : M.Tech, DESE
Manish Kumar Singh : M.Tech, DESE
Narayanan Nampoothiry : M.Tech, DESE
Rahul Chakraborty : M.Tech, DESE
III COMPONENTS
• Car chassis and wheels set including motors ,Raspberry Pi 4
• TIVA C microcontroller
• H bridge motor drivers
• Rpi cam
• Ultrasonic Sensor module HC-SR04
• IR sensors
• Power bank – for battery backup
IV IMPLEMENTATION
4.1 Device Setup
• Primary Device – Raspberry Pi 4
The first step is to set up the Raspberry Pi as the master device. This involves installing the necessary operating
system, configuring the network settings, and connecting the required peripherals.
• Secondary Device – TIVA C series microcontroller
TIVA is set up as the secondary device. The process includes connecting the board to the computer, and config uring the necessary libraries.
4.2
Communication Link Establishing a communication link between the primary and secondary devices is crucial for real-time data transfer. This is achieved by implementing a suitable protocol, such as Serial Communication or I2C, and ensuring proper synchronization between the devices.
4.3
Image processing
To enable the self-driving car to interpret and understand its surroundings, image processing techniques using OpenCV4
are to be employed. OpenCV4 provides a set of functions for tasks such as object detection, lane detection, image segmenta-
tion, feature extraction, and object tracking. These algorithms allows the car to identify and understand different elements in
its surroundings. Here RPI cam is integrated with the RPI4 which would provide image datasets to be further processed
by the device and take necessary actions.
4.4
Machine learning
Machine learning techniques needs to be incorporated into the self-driving car system to enhance its object detection and
decision-making capabilities. A cascade classifier needs to be trained to detect stop signs, traffic lights, and other objects of
interest. This involves collecting a dataset, images, and train ing the classifier using techniques like CNN.
V A PPLICATIONS
Traffic light detection and signage detection like STOPsigns
Image processing techniques need to be employed to identify and track traffic lights in real-time. Machine learning models
were trained to classify the different states of the traffic lights, such as red, green, and yellow. Similarly, different signs can
also be detected and the car can come to a halt in case of STOP sign. Here, we can use RPI cam which will continuously sam-
ple image data and the primary device can process and send certain signals to the secondary device which can further di-
rect the motor driver to take appropriate action like stopping the car, starting it etc.
code: Embedded_Mini_Project_Code
video: self driving car
Recent Comments