Umaid Asim

Autonomous cars are futuristic self-driven cars which are more efficient and therefore are able to minimize the risks of crashes. In this project, we have developed a prototype of an autonomous car. The car is able to visualize its surrounding with the help of the LIDAR sensor, IR sensor, Radar sensor, Ultrasonic sensor and camera. With the input from these devices, it is be able to create a map of its surroundings. The vehicle is able to control itself and move automatically by localizing… Show more

Autonomous cars are futuristic self-driven cars which are more efficient and therefore are able to minimize the risks of crashes. In this project, we have developed a prototype of an autonomous car. The car is able to visualize its surrounding with the help of the LIDAR sensor, IR sensor, Radar sensor, Ultrasonic sensor and camera. With the input from these devices, it is be able to create a map of its surroundings. The vehicle is able to control itself and move automatically by localizing itself and planning its path. This is the basic idea on which the Futuristic Autonomous Car is developed and designed. Although the project is done on a model car and simulation but it still provides a base for the development in this technology. The making of the Futuristic Autonomous car is setting a base for the transformation of standard vehicles to fully autonomous vehicles by implementing and adapting to various upcoming technologies. Show less

This project concentrates on the three major approaches to detect driver’s drowsiness. The eye closure detection is a first way to detect drowsiness by computing the Eye Aspect Ratio (EAR). This is a machine learning based technique which deploys HOG + SVM algorithm to detect the face of the driver and the 68-point facial landmark predictor is applied on that detected face to locate the eyes of the driver.
The other two approaches which are used to detect drowsiness are deep learning based… Show more

This project concentrates on the three major approaches to detect driver’s drowsiness. The eye closure detection is a first way to detect drowsiness by computing the Eye Aspect Ratio (EAR). This is a machine learning based technique which deploys HOG + SVM algorithm to detect the face of the driver and the 68-point facial landmark predictor is applied on that detected face to locate the eyes of the driver.
The other two approaches which are used to detect drowsiness are deep learning based that uses pre-trained VGG16 model and transfer learning is applied to create and train our own FC layers. We created two separate models and prediction classifiers one is for yawning detection and other is for head down movement detection.
These two models are concatenated with the main code of eye closure and hence a multi-threaded environment is created to run these detections in parallel. Locks are acquired to ensure thread safety.
The audio alert signal is also generated whenever drowsiness is detected using gtts and pygame mixer. The real-time data of these detections is then posted on IoT platform for analysis using http request.
Show less

By applying Deep Learning, a CNN is created to control the car autonomously by using a regression algorithm to generate the value of the steering angle via back propagation to maneuver the vehicle.

Due to an increase in the demand for automobiles, traffic hazards and road accidents are also becoming frequent. The accident is detected by collaboration of the Accelerometer sensor and vibration sensor. This would get the location information from GPS module and would send an Emergency message with accident location through GSM module to the nearest emergency center. The whole operation is controlled by Arduino UNO. This proposed system is designed to provide timely informs with minimal… Show more

Due to an increase in the demand for automobiles, traffic hazards and road accidents are also becoming frequent. The accident is detected by collaboration of the Accelerometer sensor and vibration sensor. This would get the location information from GPS module and would send an Emergency message with accident location through GSM module to the nearest emergency center. The whole operation is controlled by Arduino UNO. This proposed system is designed to provide timely informs with minimal delay due to which the percentage ratio of deaths will decrease. Show less