robotics and computer version

ROBOTICS & COMPUTER VISION
IN
SWARM INTELLIGENCE & TRAFFIC SAFETY
ABSTRACT

An automotive controller that complements the driving experience must work to avoid collisions, enforce a smooth trajectory, and deliver the vehicle to the intended destination as quickly as possible. Unfortunately, satisfying these requirements with traditional methods proves intractable at best and forces us to consider biologically -inspired techniques like Swarm Intelligence. A controller is currently being designed in a robot simulation program with the goal of implementing the system in real hardware to investigate these biologically-inspired techniques and to validate the results. This paper presents an idea that can be implemented in traffic safety by the application of Robotics & Computer Vision through Swarm Intelligence.

CONTENTS

1. Introduction

2. Motivation

3. Research

4. Impression

5. Bibliography

INTRODUCTION

We stand today at the culmination of the industrial revolution. For the last four centuries, rapid advances in science have fueled industrial society. In the twentieth century, industrialization found perhaps its greatest expression in Henry Ford's assembly line. Mass production affects almost every facet of modern life. Our food is mass produced in meat plants, commercial bakeries, and canaries. Our clothing is shipped by the ton from factories in China and Taiwan. Certainly all the amenities of our lives - our stereos, TVs, and microwave ovens - roll off assembly lines by the truck load.

Today, we're presented with another solution that hopefully will fare better than its predecessors. It goes by the name of post-industrialism, and is commonly associated with our computer technology with Robots and Artificial Intelligence.

Robots are today where computers were 25 years ago. They're huge, hulking machines that sit on factory floors, consume massive resources and can only be afforded by large corporations and governments. Then came the PC revolution of the 1980s, when computers came out of the basements and landed on the desktops. So we're on the verge of a "PR" revolution today – a Personal Robotics revolution, which will bring the robots off the factory floor and put them in our homes, on our desktops and inside our vehicles.




Figure 1: Robotic Automobile Assembly System.

What is a Robot?

Robots can take many forms-contrast Star Wars R2D2 and C3PO with the Sojourner Rover that ambled around the Martian countryside last year, and with a factory robot arm that spends 24 hours a day welding. But every robot has two attributes:

A "brain," which could be anything from a sophisticated computer down to a primitive control program Movement (either the computer itself moves, or it controls an arm or other moveable part).

There is no standardized definition, although efforts are under way to
Develop one.

A robot has three essential characteristics:

It possesses some form of mobility

It can be programmed to accomplish a large variety of tasks

After being programmed, it operates automatically





Figure 2: Robot Mine Detectors

A computer is not a robot because it lacks mobility. Special-purpose machines are not robots because they automate only a few tasks. Remote control devices work only with human participation and therefore are not robots.

The word 'robot' entered the English language in 1923 when the play 'R. U. R. (Rossum's Universal Robots)', written by the Czech author Karel Capek, was produced in London. (In Czech the word 'robota' means 'heavy labour'.)

The robot concept remained science fiction until 1961 when Unimation Inc. installed the world's first industrial robot in the US. Australia's first robot, also made by Unimation Inc., was introduced in 1974.
Up to now, most of the approximately 650,000 robots installed worldwide have been used in manufacturing. Typical applications are welding cars, spraying paint on appliances, assembling printed circuit boards, loading and unloading machines, defense, in satellite and telecommunication, surgery, and placing cartons on a pallet. The automobile and metal-manufacturing industries have been the main users. The mobility of these robots generally has been limited to a programmable mechanical arm. In some installations the platform on which the robot arm is mounted can travel automatically along a fixed rail.

The International Organization for Standardization (ISO) has developed an international standard vocabulary (ISO 8373) to describe 'manipulating industrial robots operated in a manufacturing environment'. According to this standard such a robot must possess at least three programmable axes of motion.

The International Federation of Robotics (IFR) and the Australian Robot Association follow this ISO standard when compiling robot statistics. Machines working in a manufacturing environment that have only one or two programmable axes of motion therefore are not included in these statistics.

Although the vast majority of robots today are used in factories, advances in technology are enabling robots to automate many tasks in nonmanufacturing industries such as agriculture, construction, health care, retailing and other services. Australia’s most famous robotics research project was concerned to develop a robot capable of shearing sheep.
Technologies that are being developed to extend robot capabilities include machine vision and other sensors, vehicles that can travel automatically on a variety of surfaces, and mechanisms able to manipulate flexible materials without damaging them. It is anticipated that robots will be utilized in the 21st century not only in industry but also at home. Potential domestic applications include assisting elderly or busy people to carry out tasks such as cleaning or cooking. The ISO has not yet produced a standardized definition of a robot used in non-manufacturing applications. According to the IFR such a robot is 'a machine which can be programmed to perform tasks which involve manipulative and in some cases locomotive actions under automatic control'. Robots initially have been installed in factory environments where the tasks to be done can be precisely controlled. However, it is impossible to program a robot so that it always acts correctly in an environment that is poorly understood or loosely structured. Occasional human intervention will be required to provide high-level guidance to robots working in such environments. The design of suitable human/robot interfaces is expected to become an important priority.

MOTIVATION

The goal of this project is to work toward developing a complementing automotive controller that improves upon the driving experience. The controller will monitor certain road conditions and will override the human driver only in emergency situations. When overriding, it should have three critical priorities:

Minimize propensity and severity of collisions.

No control system is perfect. It is impossible to guarantee the elimination of automobile collisions. Automobiles are complex mechanical and (increasingly) electronic systems. In the rare case that enough components fail at the same time, no amount of redundancy
can immediately restore correct operation of the vehicle. The goal of any system, given a certain cost, is to minimize the probability of a collision and, if a collision is unavoidable, lessen the severity of the impending collision.

Enforce a smooth ride.

A control system, which causes an automobile to violently weave through traffic, should be considered inferior to a system, which sends the car along a smoother, more predictable trajectory. An uncomfortable and unpredictable ride is unpleasant for the passengers and may be dangerous for other drivers on the road. A smoother ride also results in less wear and tear on the components of the automobile and prolongs the life span and reliability of critical parts.

Get the passenger from point A to point B as quickly and efficiently as possibly.

Of course, the ultimate goal of any automatic vehicular controller is to deliver the passenger to his/her intended destination. If this proves to be unrealizable (hardware fault, streets closed, etc) then the system should give the passenger the option to abort the trip or transport the passenger to a point as close as possible to the original intended
destination.

Figure 3. Junction Of Streets.

These requirements imply the necessity of introducing a priority-based architecture for the complementing controller. The controller will do all it can to deliver the passenger(s) to the destination as quickly as possible unless this results in an un smooth ride. Likewise, the controller will enforce a smooth ride unless the safety of the passenger(s) is/are threatened.
The approach I propose is the insertion of an intermediate controller situated between the human driver and the automobile actuators.
In addition to the above constraints, the automobile control system must also be able to cope with a diversity of vehicles and drivers while coping with the environment in a robust way.

RESEARCH

Satisfying all of these conditions would be a tall order for traditional control algorithms. As a result, we look for inspiration from biological systems. The Principal advantage of a biologically inspired approach is that such techniques have stood the test of eons of competition and evolution. Not only are these techniques robust, they also have the advantage of scalable and distributed operation, as well as acceptance of existing heterogeneous agents.
A specific biologically inspired approach that seems well optimized for understanding collective phenomena (like automobile traffic) is Swarm Intelligence. Swarm Intelligence provides a framework in which autonomy, emergence, and distributed robustness replace centralized control. This is analogous to comparing birds flocks to a complex man- made air- traffic control system that results in countless flight delays and lost luggage.


Figure4. Screenshot Of Webots2.0
Simulation Program.

IMPRESSION

Sample traffic situations will be simulated in the WEBOTS 2.0 simulator. The simulated automobiles are controlled by a subsumption architecture. A simplified model of a human driver (which is aware of his/her speed, orientation, and what lies within his/her field of view) will just try to avoid other cars and follow the lanes. If, for whatever reason, the simulated human driver causes the car to enter any undesirable situation, the driver will first be warned. Only when the situation becomes dire and requires immediate evasive action, will the complementing controller override the driver. In all other cases, the commands given by the driver (steering wheel angle, gas/brake pedal deflection, etc) are passed directly to the actuators.


Figure 5. Prototype Implementation Of Traffic Scenario With Real Robots.

The complementing controller will have access to data from on-board obstacle sensors and lane sensors in order to have an awareness of the state of the environment. The sensors on the Road that uses Radio signals will provide the necessary traffic information. Evolutionary techniques will be used to suggest optimal placements and configurations for the sensors on the vehicle as well as other controller parameters. Using the GPS (Globe Positioning Satellite) System the on-board computer systems gives relative location of the vehicle. The initial simulations will take place on a straight three-lane highway but curved streets may be added later (see Figure 4). Currently, the Webots simulator does not simulate the holonomicity of real automobiles. Specifically, Webots was designed to simulate small round robots with two wheels on either side. This presents a problem in regards to how a real traffic situation should be scaled so that a simulation can be realistic. An impending software revision of the simulator should resolve the issue. The Webots simulator is also not optimized for very large macroscopic simulations (hundreds of automobiles and drivers). For extremely large simulations, cellular automata-based platforms (e.g. Transims) may eventually have to be leveraged. Finally, implementing this system in real robots (see Figure 5) would provide concrete confirmation or refutation of any conclusions obtained during simulation.


CONCLUSION

The goal of this project is worked towards developing a complementing automotive controller that improves upon the driving experience. The controller will monitor certain road conditions and will override the human driver only in emergency situations. It would be a tall order for traditional control algorithms. As a result, we look for inspiration from biological systems. Swarm Intelligence provides a framework in which autonomy, emergence, and distributed robustness replace centralized control. The sensors on the Road that uses Radio signals will provide the necessary traffic information. Finally, implementing this system in real robots would provide concrete confirmation or refutation of any conclusions obtained during simulation.


BIBLIOGRAPHY

1)Swarm Intelligence: From Natural to Artificial Systems. Bonabeau, E., Dorigo, M., Theraulaz, G. New York: Oxford University Press,2001.
2)A Robust Layered Control System for a Mobile Robot. Brooks, R. IEEE Journal of Robotics and Automation, 2002.
3)Robot Herds: Group Behaviors for Systems with Significant Dynamics.
4)Proceedings of Artificial Life IV, 1999. Reynolds, C.
5)Personal Robotics by Brent Baccala.
http://www.linux-guruz.org/
http://www.robot-automation.com/robot-automation-system.htm