Teaching Cars to see A future of self driving car lane detection system using road lane edges

Project Title

Teaching Cars to see A future of self driving car lane detection system using road lane edges

Project Area of Specialization Artificial IntelligenceProject Summary

The first ever vehicle was introduced back in the 18th century. And since then we have seen tremendous amount of progress in that field. From the use of muscles to steam, from steam to fuel and now towards complete automation, we have seen our transportation system developing day by day. However, it also comes with a large number of casualties as well. Around 80 years ago, Bel Geddes in his book, Magic Motorways (1940), predicted a revolutionary development in the transportation system and also presented a huge argument that human drivers need to be removed from the driving seats of the vehicles. Now after 80 years, we are witnessing that prediction in live action which is said to be a total game changer in the field of vehicles. Established and well-known companies such as Tesla, Mercedes Benz, Nissan and Google etc. are in a tough competition with each other to be the first one to master in driverless cars. A more technological term for driverless cars is autonomous vehicles. 

Our project is about creating a prototype for an autonomous vehicle that with the help of artificial intelligence techniques and some hardware sensors will try to eliminate the need of a human driver from the roads. 
Computer Vision and Neural Networks are the two most widely developing fields in Computer Sciences and we are making use of both of them in our project to create a software system that with the given data is able to take decisions such as steering actions, lane keeping, lane changing, speed control etc. 

Project Objectives

On a bigger scale, the aim of this project is to take an initiative towards automation of vehicles in our society and decreasing the number of road casulties that take place so frequently around us. 

Particularly, our project has the following aims and objectives. 

1. To build a prototype of a small autonomous vehicle which should be able to detect the lane lines on a fabricated road by the use of a head-mounted camera.
2. To build a software system that will detect the lane lines by a given video input of a road using Image Processing System.
3. Making the prototype learn how to drive in a single lane. 
4. To make the prototype be able to navigate between the lanes based on the design and situation of the road.
5. To make the prototype be able to detect objects such as other vehicles and traffic signs by using Machine Learning.
6. To control the steering command of the prototype based on its surroundings.

Project Implementation Method

The architecture of the self driving cars is divided in two parts. 

1. The Perception System 

2. The Decision Making System 

The perception system mainly performs the following tasks. 

• self-driving-car localization
• static obstacles mapping
• moving obstacles detection and tracking
• road mapping
• traffic signalization detection and recognition 

And the decision making system has the following duties. 

• route planning
• path planning
• behavior selection
• motion planning
• control 

So the first step in the implementation is to create a Perception system. 

In which the first and most important part is to set up of Raspberry Pie Operating System. 

And then integrating a camera module with it. 

To make our prototype see objects and lanes around it, we install OpenCV which will be the eye of our car. The car will get its input from the camera and those video clips will be used in the Decision Making system to contrrol the actions of the car. 
The next step is to create a lane detection system using OpenCV in Python. The program will be able to detect lanes from the input video. 

The lane markings are used to compute the steering angle for the car according to the given lane. 

Now that our car will be able to detect lane lines, we want it to detect traffic signals and other objects around it by creating a neural network. 
The model training process is done by the following steps. 

1. mage collection and labeling (20–30 min)
2. Model selection
3. Transfer learning/model training (3–4 hours)
4. Save model output in Edge TPU format (5 min)
5. Run model inferences on Raspberry Pi 

The last and the most important part of the Decision Making System is the Motion Control. 

The car is programmed to change its speed on detection of each object. 

Benefits of the Project

Generally self driving cars have a lot benefits which are: 

1. Less Traffic 

2. Increase in Safety 

3. Lesser Road Casualties 

4. Better Transportation Services 

5. Reduced Emissions 

6. More free time 

And that is why the whole world is working towards the automation of cars. 

Our project is small contribution towards that research. 

Our projects aims to create such software models that can be deployed in any big vehcile at any time. 

Technical Details of Final Deliverable

The final deliverable of our project will be a hardware and software integrated system which we will call a prototype of a self driving car or a toy car. The hardware system will contain a toy car, a Raspberry Pi borad which will actually be the heart of our whole project, and a google's Edge TPU. And the software part is coded using Python programming language which will contain two techologies. 

1. OpenCV for lane detection and lane tracking 

2. Convolutional Neural Networks for object detection 

The steering and break control system are also in the software part as a form of python functions. 

Final Deliverable of the Project HW/SW integrated systemCore Industry TransportationOther IndustriesCore Technology Artificial Intelligence(AI)Other TechnologiesSustainable Development Goals Good Health and Well-Being for People, Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	61500
Raspberry Pi Board	Equipment	1	8000	8000
SunFounder PiCar Kit	Equipment	1	18000	18000
Google's Edge TPU	Equipment	1	12000	12000
64 GB Micro SD Card	Equipment	1	1500	1500
Batteries and battery charger	Equipment	4	3000	12000
Miscellaneous	Miscellaneous	1	10000	10000