You are on page 1of 28

MATHEMATICAL THEORY OF AUTO

DRIVER FOR AUTONOMOUS VEHICLES

AUTO 1035 Automotive Major Project

June 16, 2010


Mathematical theory of auto driver for autonomous vehicle Anand kiran

CONTENTS
1. INTRODUCTION

2. VARIOUS TYPES OF AUTONOMOUS VEHICLE

3. WORKING METHODOLOGY

4. FUTURE MODEL

5. SOCIO ECONOMICS IMPACTS

6. CONCLUSION

©2010 AUTO1035-Automotive Research Project 1


Mathematical theory of auto driver for autonomous vehicles Anand kiran

INTRODUCTION

Auto driver or auto pilot


system is a smart system
that guides the vehicle by
itself without any human
assistance.
Auto driver system reduces
the driving stress for the
driver by following the
desirable path by changing
the steering angles
automatically.

©2010 ©2010 AUTO1035-Automotive Research Project 2


Mathematical theory of auto driver for autonomous vehicles Anand kiran

VARIOUS TYPES OF AUTONOMOUS


VEHICLES

FULLY AUTOMATED
SYSTEMS

ENCHANCEMENT OF
INFRASTRUCTURE

ASSISSTANT SYSTEMS

©2010 ©2010 AUTO1035-Automotive Research Project 3


Mathematical theory of auto driver for autonomous vehicles Anand kiran

FULLY AUTOMATED SYSTEMS

• Fully autonomous systems requires a car to drive itself to a pre-set target


using unmodified infrastructure.
• These vehicles require a high level software that identifies and classifies
objects that are provided by generic sensor systems such as LIDAR,
RADAR,GPS/INS,
• The input is recorded from the sensors in a variety of formats and external
information to build an ontological structure.
• This model is effective than the remaining.

©2010
©2010 AUTO1035-Automotive Research Project 4
Mathematical theory of auto driver for autonomous vehicles Anand kiran

ENCHANCEMENT OF INFRASTRUCTURE

• This system of autonomous model works only on developed infrastructure


such as specified lanes or specified areas.
• This system is a combination of autonomous vehicles and central
supervisory system.
• The central system plans routes and autonomously guides and plans the
vehicles from A to B (pre-determined targets) and also guides the traffic in
required zones.

©2010
©2010 AUTO1035-Automotive Research Project 5
Mathematical theory of auto driver for autonomous vehicles Anand kiran

ASSISTED SYSTEMS

• This system involves minor human involvement.


• Potential fully autonomous systems are not created so far these systems can
be seen as incremental stepping stones in that direction.
• Few examples of the systems are
1. Cruise control
2.Avoidance strategy

©2010
©2010 AUTO1035-Automotive Research Project 6
Mathematical theory of auto driver for autonomous vehicles Anand kiran

WORKING METHODOLOGY

• we develop an algorithm to guide and control a vehicle on path


autonomously .

• Assuming that path is mathematical determined by equations.

• Vehicle can follow the given path only if it turns about the centre of
curvature of the path at a correct distance equal to paths radius of curvature.

©2010
©2010 AUTO1035-Automotive Research Project 7
Mathematical theory of auto driver for autonomous vehicles Anand kiran

BASIC TERMINOLOGY

FIG 1: A POSITIVE AWS STEERING VEHICLE IN LEFT TURN

Fig 1: A Positive AWS steering vehicle in left turn

©2010
©2010 AUTO1035-Automotive Research Project 8
Mathematical theory of auto driver for autonomous vehicles Anand kiran

• System uses Ackerman steering condition

• The front and rear tracks are Wf and Wr.

• The steering angles of the front inner and outer wheels are δif and δof .

• R is the kinematic radius of rotation of a vehicle.

• O is the curvature centre .

• C is the centre of mass of the car .

• l is the called the wheel base it the distance between the front and rear
axle .

• c1 and c 2 are the longitudinal distance from O

©2010
©2010 AUTO1035-Automotive Research Project 9
Mathematical theory of auto driver for autonomous vehicles Anand kiran

AUTO DRIVER ALGORITHM

• We introduce a smart algorithm to allow the wheels to change the mode of


steering depending upon the position and orientation of the vehicle and to
adjust the steer angles for various purposes.
• A smart steering system is introduced called active steering system or
steering system by wire .This steering system provides a negative steering at
low speeds and positive steering at high speeds .
• We analyse the system consider that car is moving on a given road with a
function Y=f(X) in a global coordinate frame G. The vehicle frame B
attached at the car centre of mass .
• Assuming that the z and Z axes are parallel to each other
• Assuming that the orientation angle ψ between G and B ( x and X axes).
• Xc and Yc indicates the coordinates of road curvature in curvature centre
Cc.

©2010
©2010 AUTO1035-Automotive Research Project 10
Mathematical theory of auto driver for autonomous vehicles Anand kiran

Fig 2: illustration of the car that is moving on road at point that Cc is centre of curvature

©2010
©2010 AUTO1035-Automotive Research Project 11
Mathematical theory of auto driver for autonomous vehicles Anand kiran

• RAIDUS OF THE CURVATURE OF ROAD AT GIVEN POINT

• DETERMINING THE BODY AND GLOBAL COORDINATES AT THAT


POINT

• FINDING THE KINEMATIC CHARACTERSTICS

• STEER ANGLES OF THE VEHICLE

©2010
©2010 AUTO1035-Automotive Research Project 12
Mathematical theory of auto driver for autonomous vehicles Anand kiran

• ideal motion of the vehicle on elliptic road with road radius of curvature is
plotted.

• Fig 3: the radius of curvature R for an ideal motion on elliptic road

©2010
©2010 AUTO1035-Automotive Research Project 13
Mathematical theory of auto driver for autonomous vehicles Anand kiran

VEHICLE DYANMICS

The vehicle dynamics plays a prominent role in determining the vehicle


stability. In order to achieve high accuracy in this system we introduced
POSITION ERROR
The distance between the actual centre of rotation and the curvature centre

Fig 4 : represents the position error

©2010
©2010 AUTO1035-Automotive Research Project 14
Mathematical theory of auto driver for autonomous vehicles Anand kiran

ORIENTATION ERROR
The orientation error denotes the angle error of the system

Fig 9 : angles involved in the vehicle motion ( indicates the orientation error) λ)

©2010
©2010 AUTO1035-Automotive Research Project 15
Mathematical theory of auto driver for autonomous vehicles Anand kiran

RESULTS
By incorporating position error and orientation error with adjusted steer angles
we would achieve the following results

Fig 10: adjusted front steer angles as the function of speed

©2010
©2010 AUTO1035-Automotive Research Project 16
Mathematical theory of auto driver for autonomous vehicles Anand kiran

Fig 11: Adjusted rear steer angles as a function of speed

©2010
©2010 AUTO1035-Automotive Research Project 17
Mathematical theory of auto driver for autonomous vehicles Anand kiran

Fig 12: kinematic turning radius a function of speed

©2010
©2010 AUTO1035-Automotive Research Project 18
Mathematical theory of auto driver for autonomous vehicles Anand kiran

EXPERIMENTAL MODEL
This model is created with an idea to combine the mathematical algorithm
with sophisticated hardware technology to develop a highly accurate and
responsive autonomous vehicle.

Fig 13 : basic block diagram of the system

©2010
©2010 AUTO1035-Automotive Research Project 19
Mathematical theory of auto driver for autonomous vehicles Anand kiran

THE ARCHITECTURE:
• The architecture model consists of main stages

fig 14 : model flow of data into the vehicle

• Environmental sensing: sensors that are mounted on the vehicle gives outputs and
those are used to modal the environment (roads, lanes).
• Planning: once environmental modal is confirmed, planning stage develops various
strategies, suitable scenario is chosen
• Coordination: finally the output data will be coordinated to the power train
• Decision control : the feedback errors are given to control task and final output is
given in terms of motion vector i.e. Acceleration and steering with respect to
environment.

©2010
©2010 AUTO1035-Automotive Research Project 20
Mathematical theory of auto driver for autonomous vehicles Anand kiran

HARDWARE REQUIREMENTS
• LIDAR, RADAR, GPS/INS, FLIR.
• The framework will accept inputs from sensors in a variety of formats and
external information sources such as satellite, Wi-Fi and DSRC to built the
environmental model.
• Both high-level(ultra sonic) and low-level(infrared) sensor fusion
techniques will be incorporated.
• Computer with Pentium 4 processor at 2.8 GHz and windows Xp
professional dedicated to image sensing and to run dynamic algorithms.
• Camera with high resolution of 750(width)*400(height) pixels are needed.

©2010
©2010 AUTO1035-Automotive Research Project 21
Mathematical theory of auto driver for autonomous vehicles Anand kiran

PLACEMENT OF HARDWARE ON A VEHICLE

Fig 15 : Placement of hardware on the vehicle

©2010
©2010 AUTO1035-Automotive Research Project 22
Mathematical theory of auto driver for autonomous vehicles Anand kiran

WORKING PROCEDURE FOR EXPERIMENTAL


MODEL
• Three stages in the working model
• PREDICTION: Prediction is based on the navigation system (GPS) will deliver both the
global coordinate and local coordinate of position of vehicle on the map environment.
Prediction helps in localisation of the vehicle on the street map. The latest navigation
system capable of delivering more precise information about the road curvature etc.
• PERCEPTION: By using the information given out from the prediction we can
recognize the upcoming curves and intersections can be predicted from a long distance
ahead of the vehicle.
• FUSION: Both the prediction and perception models are fused together to develop a
probalistic environment model. This model contains the position and yaw rate of the
vehicle relative to the lane. The width of the each lane and the curvature by means of a
curve radius. The obstacles are represented mainly by their positions relative to the
vehicle , their size and motion vector modelling the speed relative to the vehicle.
• The entire data obtained during this process is integrated to power train controller to
operate the vehicle.

©2010
©2010 AUTO1035-Automotive Research Project 23
Mathematical theory of auto driver for autonomous vehicles Anand kiran

SOCIO- ECONOMIC IMPACTS


1.SAFETY: These autonomous systems fall under intelligent vehicle control
system helps in reduction of accidents ,especially during lane change and
turn at intersections etc.
2. IMPACTS ON TRAFFIC: Cars in the mere future would be able to
seamlessly merge into moving traffic and exit into traffic with ease. With
reduction of traffic there is a chance in the fuel of economy.
• Speed can be eventually increased because there is no chance of human
errors.
• Parking assistants with system is made is easy .especially parking problems
in urban scenarios will be reduced subsequently.

©2010
©2010 AUTO1035-Automotive Research Project 24
Mathematical theory of auto driver for autonomous vehicles Anand kiran

CONCLUSION

• This experimental model is implementation of an idea which has a scope of


development in the future. This entire model is made to run on an ideal
scenario. With minor modification in the hardware we can test this system
in a wide range of environments such as heavy traffic scenario with traffic
lights, pesdestrian walking on the road and can be used to whilst in the urban
scenarios.
• Keeping the socio economic impacts in the mind people should trust the
working ability of the autonomous systems and they should encourage these
systems for further development.

©2010
©2010 AUTO1035-Automotive Research Project 25
Mathematical theory of auto driver for autonomous vehicles Anand kiran

THANK YOU

©2010 AUTO1035-Automotive
©2010 Research Project 26
Mathematical theory of auto driver for autonomous vehicles Anand kiran

QUESTIONS

©2010
©2010 AUTO1035-Automotive Research Project 27

You might also like