Java withrealworldtechnology

By
Anju George.K
Java Developer
Ociuz Technologies

Programming Language
• Description
• It is a set of instructions for commanding the
computer to perform some operations.
• Programmers write various programming
languages.
• Language Categories
• Machine Language
• Assembly Languages
• High-Level Language

• Machine Language
– It is a numeric language specifically
understood by a computer’s processor
– Machine language is machine dependent (not
portable).
– It makes humans cumbersome, slow, and
error-prone to develop system.

• Assembly Languages
• Represent elementary operations of computer in
English-like abbreviations
• Assemblers are developed to convert assembly to
machine language.
• It has a one-to-one relationship with machine
language.
1 0000 E8 MOV A,R0
2 0001 FA MOV R2,A
3 0002 2D ADD A,R5

• High-Level Languages
– They speed up the programming process.
– Compilers convert them to machine language
– Instructions is comprehensible to humans
• Looks mostly like everyday English
• Contain common mathematical notation
1 int sum, i, j;
2 i=100;
3 j=50;
4 sum = i + j

What is Java?
• History
– Green Project Announced by Sun in 1991
• Motivation: intelligent consumer-electronic devices
• Creator: James Gosling
• New Language: Oak
– Announcement in 1995
• Industry Conference in May 1995
• Netscape announced an agreement to integrate Java into
its browser.

• Properties
• It is a high-level programming language
• It is simple and purely object-oriented.
• It is portable, that is, executable at
different kinds of platform.
• It is developed in world wide web
applications.
• Write Once Run Anywhere (WORA).
• Designed for easy Web/Internet
applications.
• Widespread acceptance.

Characteristics of Java
• Java is
i. simple
ii. object-oriented
iii. distributed
iv. interpreted
v. robust
vi. architecture-neutral
vii. portable
viii. performance
ix. multithreaded
x. dynamic
xi. secure

Features removed from c++
 pointers to avoid unauthorized access of memory locations.
 excluded structures, unions and enum data types.
 operator over loading
 Preprocessor plays less important role in C++
and so eliminated entirely in java.
 It is not possible to declare unsigned integers in java.
 It is not possible to declare unsigned integers in java.
 In C++ objects may be passed by value or reference. In java
objects are passed by reference only.
C++ and Java

New features added in Java
 Multithreading, that allows two or more pieces of the same
program to execute concurrently.
 C++ has a set of library functions that use a common header
file. But java replaces it with its own set of API classes.
 It adds packages and interfaces.
 Java supports automatic garbage collection.
break and continue statements have been enhanced in java to
accept labels as targets.
 The use of unicode characters ensures portability.

What is Java?
• Java Virtual
Machine (JVM)
Java
source code
Java
compiler
byte-code
byte-code
JVM
EXECUTION
Java
source code
Java
compiler
byte-codeJava
source code
JVM
byte-code
JVM

• Java Class Libraries
– Java has built huge amount of libraries,
– These libraries are is known as Java APIs
(Application Programming Interface).
– Most of them are available over the Internet
and at no charge.

Class loader
( put . Class file
In memory )
Verify bytecodes &
Validate for Java’s
security violations
Translate bytecodes into a machine language
Execute the program
(display or store data values)
Edit & Store a file
( Extension .
java )
Creates bytecodes &
Stores a file
( Extension .class )

Java Development Environment
• JDK (Java Development Kit)
 Java Standard Edition (J2SE): develop client-side
standalone applications or applets.
 Java Enterprise Edition (J2EE): develop server-side
applications such as Java servlets.
 Java Micro Edition (J2ME): develop applications for
mobile devices such as cell phones

• IDE (Integrated Development Environment)
– There are number of free and commercial IDEs

What Is A jar File ?
• Java archive (jar) files are
compressed files that can store one
or many files.
• Jar files normally contain java or
class files, but other files may also be
included.
• Jar files can be run on any paltform
by double clicking jar files if the JVM
is installed.

Features of java
• Object oriented
• Platform independent
• Simple
• Robust
• Portable
• Architecturally neutral
• Multi threaded
• Distrbuted

A City Driving Robotic Car Named “Tommy
Jr.”
Paul J. Perrone
CEO – Perrone Robotics
Team Lead – Team Jefferson
TS-5925

Learn about a robotic car in a historic city driving race
and the Java™ technology that got him there
2008 JavaOneSM
Conference | java.sun.com/javaone |

Agenda
2008 JavaOneSM
• The challenge
• Tommy Jr’s parts
• Tommy Jr’s brains
• Tommy Jr’s journey
• Tommy Jr’s advancement
• Conclusions

The DARPA Grand Challenge and Tommy
2008 JavaOneSM
• 175+ mile race through desert
• No driver onboard or in remote control
• Completely Autonomous Ground Vehicles (AGVs)
• $2M prize
• Applications  Video  Site Visit  NQEs/DGC
• “Tommy” is born and advances
• 100% Java technology based bot
• Java SE platform for navigation & obstacle avoidance
• Java ME platform for controls

The DARPA Urban Challenge and Tommy Jr.
• 60+ mile through a city
• AGV city driving
•Navigation & obstacle avoidance
• Plus all rules involved with city driving
$2M/$1M/$500K prizes
•$1M grants (“Track A”) & Indies (“Track B”) Applications 
Video  Site Visit  NQEs/UCE “Tommy Jr.” is born and
advances – Track B 100% Java technology based bot
•Java SE technology & Java ME technology again
•Java technology RTS for real-time nav controls
•SunSPOT safety controller
2008 JavaOneSM

Automotive Platform: Affordable & Practical
• Scion xB
• Used 2004 = “affordable”
• Small footprint = “smaller collision surface”
• Navigable = “more responsive collision avoidance”
• Boxy = “plentiful/easier external mounting”
+ “plentiful/easier internal mounting”
• Actuation integration & drop-in
• Automatic/shifter actuator
• Cruise control drop-in
• Brake actuator
• Steering motor
• Tapped
• Throttle position
• Speedometer
• Turn signals
2008 JavaOneSM

• Embedded x86
• All of TJr’s decision making and sensor fusion
• Java SE technology & Java RTS technology
• MAX Standard & Real-Time
• Desktop-grade x86
• Front-end UI to TJr
• Stereo camera processing
• Java SE technology
• JStamp micro-controllers
• MAX Standard
• SunSPOT
• Feedback control of motors
• Java ME technology; MAX Micro
• Misc: redundant I/O controller, motor drivers, etc…
• Safety controller
• Java ME technology; MAX Micro
Hardware: COTS & Java technology capable

Hardware: COTS & Java technology capable

Sensors: COTS & Precision
• GPS – precision positioning
• Standard and differential GPS
• Hi-precision GPS – 2 to 4 inches
• INS – Inertial Navigation System – precision orientation
• Heading, pitch, roll
• GPS backup
• LADAR – laser range sensors – obstacle detection
• SICK LMS sensors – one of our strongest capabilities
• Detection ahead, on sides, and behind
• Virtual “curb feeler”
• Low hanging obstacle detection – last minute
• Stereo camera – lane detection
• Rapid integration
• Incomplete usage
• RADAR – long range warnings
• Integrated, minimal utility

Software: Layered Platform
• Java platform, Java platform, Java platform
•Java RTS API, Java SE API, Java ME API, Java SunSPOT API, Java
Comm API
• MAX Robotics Platform
•Real-Time, Standard, and Micro profiles
•General purpose rapid integration platform
•Many available drivers and libraries
• MAX Sensor Drivers
•GPS, INS, LADAR, Camera drivers
• MAX Actuation Drivers & Frameworks
•Motor & feedback controls
•Analog/digital I/O
• MAX-UGV
•UGV framework
•Extensible/configurable navigation maneuvers
•Extensible/configurable obstacle sensation
Microprocessor/Micro-controller
OS/Runtime
Java
MAX
MAX-Drivers
MAX-UGV
“Tommy Jr.”

Main Processing Loops
GPS
INS
LADAR
Camera
Obstacle
Detection
Lane
Detection
Navigation
Controls
Navigation
Maneuvers
Feedback
Controls
Safety
Controls
• Route planning (asynch)
• GPS & INS position & orientation sensing (asynch)
• LADAR sensing and obstacle detection (asynch)
• Camera sensing and lane detection (asynch)
• Navigation maneuvers (synch)
• Navigation controls (synch) Feedback controls (synch)
• Safety controls (synch)
Planning

Navigation Maneuvers
• More driving rules and complexity
• More rules for city driving (i.e. more maneuvers)
• More complex steps for maneuvers (i.e. more maneuver actions)
• Added navigation maneuver framework to MAX-UGV
• Detection of maneuvers
• Look-ahead for imminence
• Activate/de-activate, pause/resume
• Maneuver prioritization & exclusion
• Maneuver precedence/priority
• Maneuver actions
• Configurable actions
• Sequence of steps
• Navigation maneuvers
• Intersections, passing, merging, parking, zones, …

Parking Maneuver Sample
Copyright 2001-2008. Perrone Robotics, Inc. All rights reserved.
public class Parking extends ManeuverGeneric {
/** Detect imminence of maneuver and enter as appropriate **/ public boolean
detectIfImminent() {
…
Course course = super.getMovementPlanner().getCourse();
WaypointProximity nextSpot = course.getWaypoint(NextParkingSpotArrivalPointID);
startSpot = nextSpot.getWaypoint();
if(nextSpot.getDistance() < minDistanceForSpotApproach)
{ super.getMovementPlanner().setSpeed(spotApproachSpeed); if(!parking){
if(nextSpot.getDistance() < minDistanceForStartParking){ arrivalSpot = startSpot;
initParking(); // Starts parking maneuver (parking = true)
}
}
…
}

Parking Maneuver Sample – Cont’d
// Configurable parking actions
// Start approach to spot ApproachSpot
// Start slow down as approach spot (pump brakes, timeout, slow speed)
BrakePumpAction
// When within x meters of checkpoint, stop ApproachCheckpoint
// Once speed below certain mph and certain timeout, start applying reverse
WaitForCheckpointHalt
// Wait for reverse shift, then give a little gas, invert steering
WaitForReverse
// Invert steering, backup toward staging point until reached Backup
// Wait, until stop. WaitForBackupHalt
// Clear out those spot points that are part of route (i.e. make sure considered visited)
// Then proceed out of maneuver (release steering override) WaitForDrive

Parking Maneuver Sample – Cont’d
public class ApproachCheckpoint extends ParkingAction {
public boolean execute() {
if(getParking().reachedCheckpoint()){ super.getMovementPlanner().overrideSteering();
super.getMovementPlanner().setSteeringAngle(0.0);
super.getMovementPlanner().setSpeed(0.0); reachedCheckpoint = true;
return true;
}else{
reachedCheckpoint = false;
super.getMovementPlanner().setSpeed(pullInApproachSpeed); return false;
}
}

Feedback Controls
• Navigation controls
• Derived steering, speed, and shift commands
• Feedback controls
• Steering position control
• Braking level & position control
• Throttle level & position control
• Shift position control
• Closed-loop PID controls (actuator positioning)
• Speed controls (combined brake/throttle levels)

Obstacle Detection Maneuver
• More perspectives, obstacles, and degrees of freedom
• More sensor perspectives (front, sides, rear)
• Moving and static obstacles
• Multiple lanes and routes
• Embodied obstacle reactions as MAX-UGV maneuvers
• Obstacle detection
• Enhancements to LADAR-based detection
• Obstacle resolution
• Moving obstacle additions
• Had size, position, and confidence
• Add speed and direction
• Obstacle avoidance maneuvers
• Lane collision, zone collision, passing, backup, parking…

Passing Another Vehicle Sample
public class Passing extends ManeuverGeneric {
…
public boolean detectIfActive() { boolean pass = passing
|| stoppedFollowingDuringPass
|| (getMovementPlanner().getClosestUnpassableDistance() <=
closestDistanceForStopWhenUnpassable)
|| isReroutingDuringPass();
return pass;
}
public boolean detectIfImminent() { boolean immin =
(getMovementPlanner().getClosestUnpassableDistance() <=
closestDistanceForImminentPass)
|| isReroutingDuringPass();
return immin;
}
}

Steering Control Sample
public class SteeringControlPlan extends PlanGeneric
{
public void init()
{
commandSensor = (CommandSensor) super.getSensor(COMMAND_SENSOR_ID);
steerControl = (FeedbackControl)
super.getComponent(STEER_FEEDBACK_COMPONENT_ID);
steeringFeedbackSensor = (Sensor)
super.getSensor(STEERING_FEEDBACK_SENSOR_ID); steeringMotor = (Actuator)
super.getActuator(STEERING_MOTOR_ID);
steerControl.registerSensor(steeringFeedbackSensor);
steerControl.registerActuator(steeringMotor);
}
public void trigger()
{
// Get commanded steering value
SensorState state = commandSensor.getSensorState();
double steerValue = state.getDouble(DESIRED_STEERING_VALUE_ID);
// Set desired steering value
steerControl.setDesiredPosition(steerValue);
}
}

Fail-Safety
• Safety Risk = high energy + grand challenge
+ little time + little money
• Back Story
• Tommy Sr. smarting
• E-stop of little use
• Safety-critical systems knowledge
• Focus on speed controls
• Redundant speed sensing (2 speed inputs, 2 GPS speeds)
• Redundant throttle position sensing (2 inputs)
• Redundant speed calculations
• Fail-safe controls
• Watchdog safety
• Monitor processing
• Monitor sensors
• Monitor other controllers
• Cut throttle, stop car if issue

Passing Another Vehicle Sample – Cont’d
•// Initial passing impasse detection and beginning to slow down
InitialImpasseStopAction
•// Wait for car to slow down (may exit out of maneuver in meantime)
WaitForPassingSlowDownAction
•// Wait for car to come to complete halt (may exit out of maneuver)
WaitForPassingStopAction
•// Wait for an opening, timeout, and begin passing
WaitForOpeningAndPass
•// As pass, continue to check passage way, merging as needed, exit on
completion HandleExtendedMerge

Tommy
•Desert-ready robotic dune buggy
•$60K – 10 man-months software – 1 PC (Dec ’04 to Sept
’05)
•100% Java technology bot – decisions AND micro-
control
•MAX for rapid integration
•MAX-UGV for navigation & obstacle avoidance

Drop-In: Car + Actuators + Hardware + Sensors
•Four man-month (one actual month) drop-in (Jan ’07)
•Power electronics: Alternator, batteries, isolation, converters
•Actuators: braking, steering, throttle, shifting, E-brake
•Actuator electronics: Motor drivers, relays, switches
•Feedback: signals, encoders, electronics
•E-stop electronics: Commercial E-stop, government E-
•stop Sensors: GPS, INS, LADAR, Stereo Camera, RADAR
•Rapid after market drop-in approach

Base Software Platform Configuration
• One week to fully autonomous vehicle
• (Feb ’07) Java technology & MAX installs
• Basic MAX parameter configurations
• Calibration & tuning
• Autonomous navigation & obstacle avoidance
• Proven generic UGV framework with rapid
configuration

Rules Creation and Evolution
• Route planning: course maps (RNDF) + mission checkpoints
(MDF)
• Passing vehicles
• Stopping at intersections & obeying precedence order
• Queuing and following behind vehicles
• 3-point U-Turns
• Proper turn signaling
• Merging
• Open zone/lot navigation

• Parking in parking spots Lane collision avoidance
• Ground obstacle avoidance (ditches, curbs)
• Proximity collision avoidance
• Route re-planning
• GPS outage handling
• Lane detection & road following
• Low hanging obstacle avoidance
• Eight months of rule creation and testing (Mar ’07 to
Oct ’07)

Loose Ends
• Dashboard UI: manage & test TJr
• Simulation: test new rules
• Course maker: plot and load test courses
• Safety: fail-safe, don’t rely on E-stop (controlled by DARPA)
• People: engineering & operations
• Parts: sensors, car, hardware
• Facilities: basic testing, demos, large-scale tests, CA tests
• Travel: TJr legal to human drive in VA!, trailer to/from CA
• Finances/Marketing: sponsorship, logos, demos, TV, news

Applications, Sponsorships, & Kool-Aid
• March 2006: Announcement
• $2M/$1M/$500K prizes; $50K to all qualifiers, $100K to all finalists
• April 2006: Team Jefferson back
• Slam Dunk Plan ($$$): $800K
• Winning Chance Plan ($$): $400K
• Austerity Plan ($): $90K
• Summer 2006: DARPA Kick-Off
• No prize money
• Vehicles required to have stock chassis – can’t reuse Tommy Sr.
• Oct 2006: Track A awards
• $1M up front to 11 teams; mostly large teams
• Oct 2006: Track B applicants
• 89 teams total (11 Track As, 78 Track Bs)
• Interest down from 2004 & 2005
• Jan 2007: Prize adjustments
• Prize money reinstated; qualifier money not reinstated
• Jan-Feb 2007: Team Jefferson Accelerates
• Sponsorship support
• Mechanical/electrical engineering accelerates

Video Submission & Site Visit
• March 2007: Video Submission
• Demo safety, self-navigation, and passing
• 1 week SW configuration + few weeks new passing rules
testing
• 53 teams advance to site visit (11 Track As, 42 Track Bs)
• Tommy Jr. advances
• Targeting austerity plan
• June 2007: Site Visit Demonstration
• Demo safety, self-nav, passing, intersections, route planning,
queuing, U-turns, turn signaling
• Demo in front of 3 DARPA officials
• 35 teams advance to qualifications (10 Track As, 25 Track
Bs)
• Tommy Jr. aces & advances – projected among top ranked
visits
• Austerity plan goals close

Leaving for the Race
• October 2007: Leaving Virginia for California
• Tommy Jr in trailer, RV, and 3 team members
• 3 day – 2 night trip fueled by caffeine & Java technology
• Austerity goals met, partial way to winning chance target
•More testing required:
•Lane detection, route re-planning, ground obstacles
•Low-hanging obstacles?
•Completed:
• Self-navigation, route planning, lane collision avoidance,
intersections, merging, queuing, passing, zones, parking, U-
turns, signaling, proximity and zone collision avoidance, GPS
outages, fail-safety
• Development & test enhancements
• Operations & marketing enhancements

The National Qualifiers
• Day 1: Teams given course maps on USB stix w/their team
name
• Day 2: E-stop test
• Passed, DARPA presses wrong buttons
• Day 3: Circular loop with intersections and re-routing
• TJr. navigates intersections and U-turns, re-routes
• TJr clips low-hanging railway crossing arm, no E-stop issued
• Other teams w/issue, railway crossing arm ruled illegal for course
• Day 4: Circular loop with intersections and merging
• TJr. navigates intersections and obeys precedence and merging rules
• TJr. goes wide on a sharp turn in course map file
• Other teams w/issue, we re-plan for course

The National Qualifiers - Continued
Day 5: Large course representative of city
• TJr. navigates well initially
• TJr. considers re-routing, course map points clip K-barriers
• Tjr. restart, begins re-route, clips barrier, sensor damage, no E-stop
Day 6: Second run of loop w/intersections & re-routing
• Team prepared for low-hanging obstacle, but crossing arm removed
• Sensor damage causes TJr to detect ground and re-route
• Team goes off to prep for Day 7
• Gets call back early to run another course, team inquiry
• Team gets call from DARPA director that is disqualified, team appeal
Day 7: Second run of loop w/intersections & merging
• Team prepared to complete its second runs
• DARPA director denies appeal
• Urban Challenge Program Manager tries to help, decision is final

The Final Event
•11teams advance to final event (7 Track As, 4 Track Bs)
• DARPA director selects the teams, claims safety basis
•Six teams complete the course (5 Track As, 1 Track B)
• Prize Winners:
• 1st Place: Carnegie Mellon – General Motors team (Track A)
• 2nd Place: Stanford Univ – Volkswagen team (Track A)
• 3rd Place: Virginia Tech – CAT team (Track A)

Tommy Jr. Video
Perrone Robotics, Inc.

Reflections on the Journey
• What worked well
• Extraordinarily large function base generated per unit time
• Building on Java technology and a framework definitely way to go
• What didn’t work well
• Gaps in lane detection & route re-planning – time runs out
• Issues with race operations
• Rules incomplete and not-specified
• Decisions made by director vs committee
• E-stop safety not employed
• Lessons learned from race
• Design for generalized vs specific rules ($$$)
• Design for dynamic vs specified rules ($$$)
• Design for more fail-safety scenarios paramount ($$)

Reaping the Rewards
• Designs for use beyond challenge
• High profile race & technology
• Have generalized Java technology based robotics & automation
platform
• Have generalized Java technology based UGV platform
• Awareness generated
• TV, news, trade
• Hi-profile demonstrations
• Sponsors benefit before, during, & after race
• Applying to the real-world
• Distributed sensor applications for automation
• Platform and framework licensing
• Commercial UGV applications
• Vendor/sponsor technology used and licensed

Tommy Junior’s Sponsors
• Sun Microsystems – Java technology, advisors, financial
Perrone Robotics – MAX, software/systems/operations,
financial
• University of Virginia – mechanical/electrical
• Northrop
• Grumman - financial
• AutoTrader - financial
• WowWee Robotics - financial, toy bots
• BDMetrics – financial
• SICK - lasers
• 45Fix - financial
• TDFund – financial

Java withrealworldtechnology

Recomendados

Recomendados

Mais conteúdo relacionado

Semelhante a Java withrealworldtechnology

Semelhante a Java withrealworldtechnology (20)

Mais de Anju Kanjirathingal

Mais de Anju Kanjirathingal (9)

Último

Último (20)

Java withrealworldtechnology