You are on page 1of 8

ROBOTICS PROJECT REVIEW-1

TITLE:-AUTONOMOUS CAR SIMULATION

TEAM MEMBERS:-

Prishita Ray- 17BCE2405

Mahima Gupta-18BCE2128

Soorya Ram Shimgekar-18BCE0476

ABSTRACT:-
With the advent of technology to automate daily tasks, we are developing robots that
can make our lives easier. One such application that has garnered much attention
recently is the application of robotics to self driving cars. A self driving car is an
autonomous mobile robot that is capable of performing actions based on its external
environment such as avoiding obstacles and by monitoring various internal factors such
as fuel capacity, speedometer values etc. Existing solutions for this problem make use
of computer vision to generate incomplete paths and helps calculate the distance from
mean level of movement that helps detect major deviations from the path being
followed. It also makes use of proximity sensors to check nearby objects, speed
sensors to check for rash driving and actuators or motors to control brake and
accelerator movements, gear shifts etc. Since it is a continuous learning process, it will
have to train on new data in real time. The problem with the current techniques is that
they are not completely reliable and the system may be prone to failure at any point of
time, risking the passengers’ lives as well as those of others, due to insufficient training
and lesser test situations considered before dispatching. In this project, we will be using
deep neural networks and convolutional neural networks to clone driving behavior. We
will train, validate and test a model using Keras. The model will output a steering angle
to an autonomous vehicle.
We will use a simulator where we can steer a car around a track for data collection. The
retrieved image data and steering angles will be used to train a neural network and then
this model will be used to drive the car autonomously around the track.
INTRODUCTION:-
An autonomous car is a vehicle capable of feeling its environment and operating without
human involvement. A human passenger is not required to control the vehicle at any
time, nor is a human passenger required to be present in the vehicle. It can go
anywhere a traditional car goes and does everything that an experienced human driver
does. Driverless cars are much needed in the modern world in order to automate the
process of driving, and making it a much easier and smoother experience. There are
some important advantages to this technology.

Landscapes are limitless to improve convenience and quality. The elderly and the
physically disabled will have freedom. Moreover, it saves a significant amount of time, if
the car could do tasks such as dropping off or picking up someone on its own, without
having a human to waste time on such trivial tasks. They also have the ability to
dramatically reduce CO2 emissions. In a recent study, experts have identified three
trends that, if adopted concurrently, will achieve the full potential of autonomous cars:
vehicle automation, vehicle electrification, and rideability. By 2050, these may be "three
revolutions in urban transport": that will reduce urban CO2 emissions by 80%
worldwide. This technology can lead to reduction in traffic congestion to as much as
30% less vehicles on the road, 40% reduction in transportation costs (in terms of
vehicles, fuel and infrastructure), improved portability and free parking lots for other
uses (school, park, community centre).

Though the technology used in the driverless cars is cutting edge but still a lot of work
has to be done for it to completely legalized.The main concerns include unpredictable
pedestrians, hacking and road infrastructures.

PROBLEM IDENTIFICATION AND JUSTIFICATION


When designing a self-driving car or an autonomous vehicle, several factors have to be
taken into consideration and corner test cases must be understood for it to work
efficiently. There are reasons to be very cautious about new risks associated with these
technologies, such as malicious abuse.

● Hackers can easily get access to the training code of the vehicle, and alter it, that
can lead to disastrous results.
● Optimistic predictions generally refer to limited self-driving ability, such as
operating in fair weather on grade-separated highways: no current technologies
allow reliable operation in heavy rain or snow, or on unpaved roads.
● They will initially be expensive, adding thousands of dollars in annual costs for
additional equipment, maintenance and specialized mapping services.
● Autonomous vehicles will need to identify and respond to a lot of obstacles,
sometimes unknown, such as children, animals, and other objects.
● The speed of driving will also depend on the weather conditions, for example it
must move at a slower pace during rain.
● Mis-categorization of potential danger signals, each situation will present new
problems and require new software codes that may interact unpredictably with
other driving instructions.
● Because of the diversity of possible interactions, driving in traffic requires far
more decisions, and is therefore orders of magnitude more complex software.
● Autonomous driving technologies will almost certainly cause accidents due to
unexpected conditions, probably including dramatic crashes that harm other road
users.
● Driving very close together to reduce road space requirements and wind
resistance, but this is only feasible if autonomous vehicles have dedicated lanes.

Taking into consideration all these factors, we will simulate an autonomous driving car,
that will be trained to drive efficiently within a limited boundary, such that if any foreign
object is detected along with the car in its boundary area, the emergency software
module is called, which then classifies the type of danger and takes suitable steps, to
come to a safe position. The car will be trained to identify a small number of obstacles,
but with many samples of each to improve its efficiency of movement. Moreover, we will
try to help it identify current weather conditions based on temperature, air pressure,
wind speed and humidity readings, such that it can alter its speed accordingly.

LITERATURE REVIEW:-

1. Fernandes et al., Intelligent Robotic Car for Autonomous Navigation: Platform and
System Architecture. Proceedings - 2012 2nd Brazilian Conference on Critical
Embedded Systems, CBSEC 2012. 10.1109/CBSEC.2012.26.

This paper presents the platform and system architecture of an intelligent vehicle,
presenting the control system modules allowing the vehicle to navigate autonomously.
The authors have developed works on autonomous navigation and driver assistance
systems, using CaRINA I platform to experiments and validation which includes
mechanical vehicle adaptations and the development of an embedded software
architecture, and its practical implementation. It also addresses details concerning the
sensing and acting infrastructure. Several experimental tests have been carried out to
evaluate both platform and proposed algorithms.

2. Autonomous Vehicles: Theoretical and Practical Challenges, Margarita et al., XIII


Conference on Transport Engineering, CIT2018
Autonomous driving is expected to revolutionize road traffic attenuating current
externalities, especially accidents and congestion. Carmakers, researchers and
administrations have been working on autonomous driving for years and
significant progress has been made. However, the doubts and challenges to overcome
are still huge, as the implementation of an autonomous driving environment
encompasses not only complex automotive technology, but also human behavior,
ethics, traffic management strategies, policies, liability, etc. As a result, carmakers do
not expect to commercially launch fully driverless vehicles in the short
-term. From the technical perspective, the unequivocal detection of obstacles at high
speeds and long distances is one of the greatest difficulties to face. Regarding traffic
management strategies, all approaches share the vision that vehicles should behave
cooperatively. General V2V cooperation and platooning are options being
discussed, both with multiple variants. Various strategies, built from different
standpoints, are being designed and validated using simulation. Besides, legal issues
have already been arisen in the context of highly-automated driving. They range
from the need for special driving licenses to much more intricate topics like
liability in the event of an accident or privacy issues. All these legal and ethical
concerns could hinder the spread of autonomous vehicles once technologically
feasible. This paper provides an overview of the current state of the art in the
key aspects of autonomous driving. Based on the information received in situ
from top research centers in the field and on a literature review, authors highlight
the most important advances and findings reached so far, discuss different
approaches regarding autonomous traffic and propose a framework for future research.

3.Qudsia Memon ; Muzamil Ahmed ; Shahzeb Ali ; Azam Rafique Memon ;


Wajiha Shah et al., Self-driving and driver relaxing vehicle,Proceedings-2016
2nd International Conference on Robotics and Artificial Intelligence
(ICRAI),DOI: 10.1109/ICRAI.2016.7791248.

This paper the writer got centered on 2 applications of an automatic


automobile, one during which 2 vehicles have same destination and one is
aware of the route, wherever alternative don't. the subsequent vehicle can
follow the target (i.e. Front) vehicle mechanically. the opposite application is
machine-controlled driving throughout the serious tie up, thence reposeful
driver from endlessly pushing brake, accelerator or clutch. the most plan
behind the project is,defining the one facet here into consideration is creating
the destination dynamic. this could be done by a vehicle mechanically
following the destination of another vehicle. Since taking intelligent selections
within the traffic is additionally a difficulty for the machine-controlled vehicle so
this facet has been also into consideration during this paper.
4. Shih-Chieh et al., The Architectural Implications of Autonomous Driving: Constraints
and Acceleration, Proceedings of ASPLOS’18, Association of Computing Machinery

In this paper, the authors present and formalize the design constraints for building an
autonomous driving system in terms of performance, predictability, storage, thermal and
power. They then build an end-to-end autonomous driving system using state-of-the-art
award-winning algorithms to understand the design trade-offs for building such systems.
In their real-system characterization, they identify three computational bottlenecks,
which conventional multicore CPUs are incapable of processing under the identified
design constraints. To meet these constraints, they accelerate these algorithms using
three accelerator platforms including GPUs, FPGAs and ASICs, which can reduce the
tail latency of the system by 169×, 10×, and 93× respectively. With accelerator-based
designs, they were able to build an end-to-end autonomous driving system that met all
the design constraints, and explored the trade-offs among performance, power and the
higher accuracy enabled by higher resolution cameras.

They concluded that while power hungry accelerators like GPUs could predictably
deliver the computation at low latency, their high power consumption, further magnified
by the cooling load to meet the thermal constraints, could significantly degrade the
driving range and fuel efficiency of the vehicle. It was also demonstrated that
computational capability was the bottleneck that prevented them from benefiting from
the higher system accuracy enabled by higher resolution cameras.

5. T. Banerjee ; S. Bose ; A. Chakraborty ; T. Samadder ; Bhaskar Kumar ; T. K. Rana


et al., Self driving cars: A peep into the future, Proceedings-2017 8th Annual Industrial
Automation and Electromechanical Engineering Conference (IEMECON),DOI-
10.1109/IEMECON.2017.8079556

The paper includes a unique embedded controller design of a driverless, green energy
powered, collision protected and GSM destination guided vehicle.A GPS module
accurately tracks the placement of the automotive, supply and destination, and mapping
the coordinates provides navigation Speeds of the vehicle is mechanically controlled by
keeping a secure distance, that could be a operate of speed, with the vehicle before.
Distance of the front and aspect vehicle are ceaselessly monitored by a stepper motor
controlled rotating distance measure device and also the regulation furthermore as track
dynamic are done consequently. It additionally prevents collision thanks to Associate in
Nursing obstacle. Camera with image process unit has been accustomed sense light
and traffic density. The driverless automotive is steam-powered by inexperienced
energy from a upper side star PV panel with battery backup. Helpful audio info is also
provided to the traveler.
6.Robotics and Autonomous Driving Hannah Kerner, Alan Kuntz, Jeffrey Ichnowski, and
Michael North, May 4, 2015

An autonomous vehicle design like many cyber physical systems is by its nature a
highly multidisciplinary project. The challenges associated with a subset of the overall
system, robotics, are themselves nontrivial. Any such vehicle which hopes to interact
with the world must be able to perceive, analyze, and react to its surroundings with real
time guarantees. Vehicle localization, task planning, real time motion planning, and
equipment and platform choices are integral pieces of such a system. By examining
some relevant approaches to these problems insight can be gained into the types of
computation and the equipment that may be needed in a novel and sustainable
autonomous vehicle design.

7.Lex Fridman, Daniel E. Brown, Michael Glazer, William Angell, Spencer Dodd,
Benedikt Jenik,Jack Terwilliger, Aleksandr Patsekin, Julia Kindelsberger, Li Ding, Sean
Seaman, Alea Mehler,Andrew Sipperley, Anthony Pettinato, Bobbie Seppelt, Linda
Angell, Bruce Mehler, Bryan Reime, MIT Advanced Vehicle Technology Study:Large-
Scale Naturalistic Driving Study of Driver Behavior and Interaction with Automation

The application of state-of-the-art embedded system programming, software


engineering, data processing, distributed computing, computer vision and deep learning
techniques to the collection and analysis of large-scale naturalistic driving data in the
MIT-AVT study seeks to break new ground in offering insights into how human and
autonomous vehicles interact in the rapidly changing transportation system. This work
presents the methodology behind the MIT-AVT study which aims to define and inspire
the next generation of naturalistic driving studies. To date, the dataset includes 122
participants,15,610 days of participation, 511,638 miles, and 7.1 billion video frames.
Statistics about the size and scope of the MIT-AVT dataset are updated regularly on
https://hcai.mit.edu/avt.

8.Perception, Planning, Control, and Coordination for Autonomous Vehicles


Scott Drew Pendleton 1,*, Hans Andersen 1, Xinxin Du 2, Xiaotong Shen 2, Malika
Meghjani 2 You Hong Eng 2, Daniela Rus 3 and Marcelo H. Ang Jr. 1

Autonomous vehicles are expected to play a key role in the future of urban
transportation systems, as they offer potential for additional safety, increased
productivity, greater accessibility,better road efficiency, and positive impact on the
environment. Research in autonomous systems has seen dramatic advances in recent
years, due to the increases in available computing power and reduced cost in sensing
and computing technologies, resulting in maturing technological readiness level of fully
autonomous vehicles. The objective of this paper is to provide a general overview of the
recent developments in the realm of autonomous vehicle software systems.
Fundamental components of autonomous vehicle software are reviewed, and recent
developments in each area are discussed. Based on the surveys and reviews done by
the authors, they have identified recent advancements in the field of SLAM that has
contributed significantly to the localization capabilities of autonomous vehicles. Another
active research topic in the field of vehicle mapping found by them is long-term mapping
to update the maps with static, topometric, activity and semantic data over time in order
to ensure that the vehicle can localize itself precisely and consistently with respect to its
environment.

They also conclude that recent research is progressing toward better inclusion of robot
differential motion constraints and efficient strategies for knowledge retention between
subsequent iterations of replanning. Ensuring that the autonomous system robustly
follows the intention of higher level decision making processes is crucial. Model
Predictive Control (MPC) based techniques have been an active research topic in this
area, due to their flexibility and performance. Computational time is essential in real
time applications, and therefore model selection and MPC problem formulation varies
from one application to another. It has been shown that vehicle cooperation can enable
better performance in perception and planning modules, however there is much room
for advancement to improve the scalability of multi-vehicle cooperative algorithms.
Furthermore, they state that although hardware is being standardized for V2V
communications, no standard yet exists for what information content should be passed
between vehicles.

9.Li Li ; Wu-Ling Huang ; Yuehu Liu ; Nan-Ning Zheng ; Fei-Yue Wang et al.,Intelligence
Testing for Autonomous Vehicles: A New Approach,Proceeding-IEEE Transactions on
Intelligent Vehicles ( Volume: 1 , Issue: 2 , June 2016 ),DOI: 10.1109/TIV.2016.2608003

In this paper, we have a tendency to study a way to check the intelligence of associate
autonomous vehicle. Comprehensive testing is crucial to each vehicle manufacturers
and customers. Existing testing approaches will be classified into 2 kinds: scenario-
based testing and functionality-based testing. we have a tendency to initial discuss the
shortcomings of those 2 sorts of approaches, and so propose a replacement testing
framework to mix the advantages of them. supported the new linguistics diagram
definition for the intelligence of autonomous vehicles, we have a tendency to make a
case for a way to style a task for autonomous vehicle checking and the way to gauge
test results. Experiments show that this new approach provides a quantitative thanks to
check the intelligence of associate autonomous vehicle.
PROGRESS IN WORK:
We have currently researched on existing technology available for autonomous cars
and come with our own approach to minimize inherent drawbacks in the training
process. This has helped us decide the modules that we will include within our code
base such as a script used to create and train the model using deep neural networks,
script to drive the car, module to identify weather conditions, data extraction module,
boundary restrictions module, speed control module and a simulator to visualize the
car’s behaviour in real-time situations. We will be using python for our code base. The
module descriptions have been completed. Our next step in the implementation would
be to generate the neural network code, and train the model, followed by testing it using
the simulator.

You might also like