Autonomous robotic car on an urban track

Project Title

Autonomous robotic car on an urban track

Project Area of Specialization Artificial IntelligenceProject Summary

An autonomous car is capable of sensing its environment and moving safely with little or no human input.  We are designing a prototype of a small town that includes buildings, parks and track. The track contains 2lane main roads, roundabout, one and two way roads, electronic systems(traffic lights, street lights).Along with a track we are designing an autonomous robotic car whose primary objective is to detect collision in order to drive safely and efficiently through computer vision and other sensors and to implement braking algorithm.

The input to the system is sensor measurement and the output is the steering response which either stops or turns depending on the object that is Infront of the vehicle. by running the camera feed through an image recognition network, the system is able to tell what type of objects are present and using some sensors find the distance to them. The final product will be an autonomous car that avoid collision with obstacles and also can identify specific objects and react in different ways depending on the type of object.

Project Objectives

The objective of our project is to develop a collision and braking algorithm of an autonomous car, for that we will be required to implement the following modules.

1. To design a track similar to a real world town but on a small scale that covers basic features that are present in almost every town.
2. To design an autonomous robotic car with following functions:

• To detect obsticles and their time to collision.
• To determine whether to apply brakes or not.
• To calculate safe distance between car and the obstacles
• To determine the speed of an autonomous car.

Project Implementation Method

We have three sub-modules in our project, as follows

• Track Design:

We are implementing an autonomous robotic car application on an urban track(Scaled down version of real world town) with all the main city road elements like signalized intersection, streets (1 way,2 way road), zebra crossing, buildings and parks, electronic system (traffic signals, street lights).

• Robotic car:

We are going to follow f the following strategy ...

Microprocessor based (Jetson nano):

All the sensors including camera now directly attached with a microprocessor and all the computations (AI algorithm + sensor data) are also performed on it.

• AI based collision avoidance and braking algorithm:

First, we will collect data on our track with the help of a camera that is on our robotic car. After that we will train the data set as per requirement of our project and test it on a virtual simulator before implementing it on actual hardware.

We have multiple cases of collision that we will take under consideration, few of them are listed below:

1. If pedestrian comes in-front of the car, it will calculate whether the pedestrian is static or moving. If the pedestrian in static then it will think whether to apply brake or change the lane to avoid collision. If the pedestrian is moving then there are two possibilities, our car might slow down its speed or might change its lane
2. If some other car is in-front of our self-driving car so it will calculate whether the car is static or moving. If the car is static so it will change the lane or apply brake. If the car is moving then there are two possibilities, our car might slow down its speed or might change its lane.

Benefits of the Project

According to the research the number of accidents that were reported in the year 2018-19 in Pakistan is around 10800 out of which fatal accidents were 4900 and the number of people died were around 5600 and the number of injured were 13200 and vehicles that were involved in the accident were around 13000. It is estimated that self-driving cars could decrease accident fatalities by as much as 90 percent because accidents occur due to the decisions take by the humans and humans have emotions so emotions can lead to accidents most of the time. But machines don’t have emotions so this is the main reason we will be having less accidents.

Another benefit of commercializing autonomous car is that disabled and unskilled people will get ease in this account. Unskilled people will not be now worried for learning how to drive and also the disabled people will be highly benefited with this technology.

This technology will also help the world in the delivery of things. Most of the delivery systems will automatically become better with our autonomous car

Technical Details of Final Deliverable

Our car will be moving on the track and the moment it catches any object in the eye of camera it will start to think on the basis of the data gather from the sensors and the camera itself, whether to deviate or to slow down its speed. This will be calculated by estimating the probability of each lane then the lane with the higher probability will be the target lane, this whole process is completed through some recurrent neural network. After predicting the lane sequence object’s trajectory is predicted. For a single we have multiple trajectories so we first eliminate the trajectory that is physically impossible to be executed so the decision is made on the basis of vehicle’s velocity and acceleration.
to execute the upper mentioned mechanism, we are required to have a controller that will manage the motors of our car in a way that it will be executing the mechanism properly, for that we will be using PID controller which is a simple controller and aa it needs to know how far we have deviated from target trajectory. Here P is responsible for pulling back the vehicle to the target trajectory as soon as it deviates. The D term is responsible for minimizing how quickly the controller output changes and the I term of the controller penalizes the accumulated error of the system.

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

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	71260
Jetson nano Microprocessor	Equipment	1	27000	27000
Arduino Mega	Equipment	3	1400	4200
motor driver L293d	Equipment	1	300	300
DC Plastic gear motors with wheels	Equipment	4	170	680
MPU 9250 motion sensor	Equipment	2	680	1360
Sharp IR Sensor	Equipment	2	850	1700
Ultrasonic sensor	Equipment	4	200	800
Robotic chassis	Equipment	1	500	500
LIPO	Equipment	1	2100	2100
Traffic regulation unit	Equipment	1	1500	1500
Jumper wires	Equipment	60	4	240
Wifi shield	Equipment	2	440	880
Wifi Camera(Rear+ front))	Equipment	2	10000	20000
Track Mat printing+ Building material	Miscellaneous	1	10000	10000