Self Driving Car using AI

Project Title

Self Driving Car using AI

Project Area of Specialization Artificial IntelligenceProject Summary

Autonomous cars rely on sensors, actuators, complex algorithms, machine learning systems, and powerful processors to execute software.

Autonomous cars create and maintain a map of their surroundings based on a variety of sensors situated in different parts of the vehicle. Radar sensors monitor the position of nearby vehicles. Video cameras detect traffic lights, read road signs, track other vehicles, and look for pedestrians. Lidar (light detection and ranging) sensors bounce pulses of light off the car’s surroundings to measure distances, detect road edges, and identify lane markings. Ultrasonic sensors in the wheels detect curbs and other vehicles when parking.

Sophisticated software then processes all this sensory input, plots a path, and sends instructions to the car’s actuators, which control acceleration, braking, and steering. Hard-coded rules, obstacle avoidance algorithms, predictive modeling, and object recognition help the software follow traffic rules and navigate obstacles.

There will be an agent (a robot toy car) which will maneuver in a city environment build on small scaled hardware level by avoiding obstacles and following limited traffic rules such as traffic signals and reach to end destination .

Project Objectives

After careful examination and discussing everything with the respective persons including our supervisor and various others, We came up with the following points 

• Our training agent (car) will travel on a straight path or some already guided path, it won't be a complex path 

• Car will follow traffic signals 

• It will avoid obstacles(other cars)

Project Implementation Method

After studying various research papers and contacting various research personals, what we came up is as follows 

• First of all designing a map fair enough suitable for our project objectives 

• Map will be made in a third party software called roadrunner from mathworks as we have to make a custom map 

• Importing that map into a simulation software and in our case we chose Carla, that is specifically designed for the purpose of training AI cars. 

• Running our model in that simulation environment through several days and weeks of training.

• Choosing a neural network for training our model. 

• Projecting our model map into the physical world using either lego blocks or something else that will make the buildings easily recognisable.

• A vehicle that will be our model agent for implementation in the real world.

• Just like in simulation, we will now train our agent in the physical world as well in our model map.

• After a successful training percentile Finally testing our agent into some other unknown environment.

Benefits of the Project

• Over 80 percent of all car crashes are because of human error 

• A computer never gets “distracted”, it just remains on task.

• self-driving cars decrease car accident possibilities

• Not everyone can afford a driver 

• improve traffic and congestion, especially in cities. In high traffic

• reduce gas usage and smooth out commutes.

• This new industry could create new jobs, employment opportunities and economic growth.  

Technical Details of Final Deliverable

• For the software part:

• AI algorithm that will be used is CNN(convolution neural network) to achieve the above mention objective 

• Tensorflow library will be used for AI tools 

• High end GPU requiring Software (CARLA) ,used to to acquire data and simulate the AI model on a simulated environment that (resemble / exactly ditto) to a physical model city  

• Software will be run on a high end pc that require a minimum of 4gb GPU 

• For the hardware part:

• A model city will be prepared (resemble / exactly ditto) to that one trained on the simulation and will be scaled down to toy level 

• The model city will have roods,working traffic lights ,symbols , directed cars to introduce traffic,obstacles 

• Agent:

• The AI Car will consist of the following hardware 

• Camera mount (for real time image input)

• controller~Nvidia jetson-Nano (for image process and run AI algorithm )

• Controller will also produce the necessary steering ,brake and accelerator  commands /signals

Final Deliverable of the Project HW/SW integrated systemCore Industry TransportationOther Industries Others Core Technology Artificial Intelligence(AI)Other Technologies OthersSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
Nvidia jetson nano	Equipment	1	16000	16000
Camera	Equipment	1	4000	4000
Toy Car	Miscellaneous	1	5000	5000
Map Prototype Hardware	Miscellaneous	1	5000	5000
Dedicated gpu for CARLA Software	Equipment	1	50000	50000