Self Driving cars aren’t easy to build but these students from NED created a prototype which as per the video, looks pretty cool for a final year project. The students have used a Jetson Nano microprocessor for prediction of steering angles which are required to manoeuvre the car around objects and the same microprocessor also does object detection and traffic light classification.
Object detection, in this case, is used to detect pedestrians on the road and stop the car in time. The same can be used to detect stop signs on the road and stop the car in time. The students trained the machine learning steering model around 100 laps and 36K frames, with 12 MSE and 95% r-squared. As per the students, the car is capable of working in different lighting conditions as well.
While the car still needs years more work, we can see that there is huge potential in students of Pakistan if they are capable of making a self-driving car for their Final Year Projects of bachelors. All they need now is industry support to improve the car’s functionality and features. If they are successful in creating a good enough prototype, the same industry partner can help in commercialization of the car.