Lego Robotics Kickoff Meeting

albanianboneyardAI and Robotics

Nov 2, 2013 (3 years and 9 months ago)

79 views

Case Western Reserve University
Lego Robotics

Kickoff Meeting

What’s up with his foot?


I had surgery to have some bone spurs
removed.


It happened last Friday.


It’s no big deal I’ll be wearing shoes again by
the end of the month.

Who am I?


Now that that’s out of the way…


Amaury Rolin


Undergrad In Mechanical Engineering at Carnegie
Mellon University, Pittsburgh


Worked for 4 years at National Instruments in
Austin, TX


Came to Case in 2005 to get my masters in EECS


Lead the software effort on Case’s Urban
Challenge entry in 2007

Dr. Wyatt S. Newman


My advisor


Also Mech E turned Electrical Engineer


Managed Case’s Urban Challenge team

Agenda


Intro To Case Western Reserve University and
Urban Challenge. (20 min)


LabVIEW distribution


Intro to our competition and the rules. (20
min)


Intro to LabVIEW workshop Pt I (40 min)


Field trip to other robots & Kit Distribution (20
min)


Intro to LabVIEW workshop Pt II (…)



Urban whowhat?


Urban Challenge


Congress wants 1/3 of military vehicles to be
unmanned by 2015


We’ve got a long way to go


DARPA gets research for cheap by convincing
people to compete to build self driving cars.

Who competes?


Mostly universities that have been researching
how to build robots for a long time.


Some companies that want to show that they
know the robot game so when the time comes
when the army is looking for someone that can
build their robots….


Car companies that want to add all sorts of safety
features to their cars. (mostly partnered with
universities)


Anyone who can and wants to play with really
expensive toys ($3M
-
$300K).

And you’re telling me this why?


Since it’s such an important event in robotics
and represents the bleeding edge of mobile
robotics we decided to model this years Lego
competition after it.


So stay tuned!

What do you get?


1
st

Place 2M & a giant eagle shaped trophy


2
nd

Place 1M & a big eagle shaped trophy


3
rd

Place 500K & an eagle shaped trophy


After that…a warm satisfied feeling that you
are not working on that robot any more!

History of the Urban Challenge


Used to be the Grand Challenge


Started in 2004 as a 300mi robot race through
the desert.


Winner: Carnegie Mellon University


Distance Traveled: 7mi


Conclusion: What a shambles! We better try
that again.

Next Round 2005


A little easier: 180mi race


5 Teams finish.


Stanford gets 1
st

place


One robot starts out with a tremendous lead
and 80min into the race seems like the clear
winner…until disaster strikes its name:
DEXTER

Who is DEXTER and why do we care?


DEXTER was built by the ENSCO corporation to
compete in the 2005 Grand Challenge.


DEXTER is a robot built on a dune buggy
chassis.


DEXTER is a desert eating machine.


DEXTER was Case’s adoptee for running in the
2007 Urban Challenge


DEXTER movie

Grand/Urban Challenge


Urban Challenge is a Grand challenge


Desert Challenge


When a robot caught up to an other robot the lead robot would be
paused and the other robot could pass.


Only one path and the goal was to get to the end.


Challenge was mostly terrain. Where to go was marked with waypoints
every 100ft in an RDDF file.


First robot to cross the finish line wins.


Urban Challenge


Robots have to interact with other robots and live traffic.


A network of roads is defined. Robots must be able to plan their way
through the network to complete a mission.


Robots must obey the rules of the road. Stop at intersections and yield
to traffic.


Robots are scored based on checkpoints and how safe they
appear.

So the robot has to be able to
see stop signs and lights?


No.


There are not stop lights in Urban Challenge
World.


Whether to stop at an intersection depends
on how it is defined in the RNDF (I’ll get to
that in a bit)


Where to stop was defined by the coordinates
in the RNDF.

So you drove this robot on real roads?


Sort of


We practiced on closed roads or on the
campus quad. DEXTER is definitely not street
legal.


The actual competition was on George Air
Force Base in Victorville, California

Basic Navigation


Vehicle is in autonomous mode and ready to begin run less than 5 minutes after
receipt of the MDF from DARPA.


Vehicle front bumper passes over each checkpoint in DARPA MDF in the correct
lane and the correct sequence.


Vehicle stays in travel lanes at all times unless exiting lane to avoid obstacle.


Vehicle always stops so front bumper is within 1 meter of stop line at intersection.


Vehicle always exhibits less than 10
-
second delay before proceeding at clear
intersection.


Vehicle exhibits safe behavior at all times to avoid collisions and near
-
collisions as
judged by DARPA.


Vehicle demonstrates ability to leave lane, pass a stopped car or obstacle, and
return directly to travel lane. Complete maneuver takes place within 40 meters.


Vehicle maintains a minimum safety separation of 8 meters fore and aft when
executing a passing maneuver.


Vehicle speed conforms to limits set in DARPA MDF.


Basic Traffic


Vehicle meets all criteria for navigation test.


Vehicle exhibits proper precedence order at every
intersection and does not proceed out of turn.


Vehicle never comes closer than 15 meters when
following a moving lead vehicle traveling at 15

mph on
an urban course with 20 mph speed limit.


Vehicle stays within 40 meters when following a
moving lead vehicle traveling at 15 mph on an urban
course with 20 mph speed limit.


Vehicle stops between 5 and 10 meters behind a
stopped lead vehicle.


Advanced Navigation


Vehicle exhibits correct parking lot behavior, including ability to pull
forward into and reverse out of specified parking spot without
collision and with less than 10 seconds of excess delay.


Vehicle demonstrates ability to negotiate obstacle field safely and
effectively, with no collisions and with less than 10 seconds of
excess delay.


Vehicle conducts maneuvers necessary to achieve objective
checkpoints, including U
-
turns and route re
-
planning when roads
are blocked. A U
-
turn may be effected through one or more three
-
point turns.


Vehicle navigates roads with sparse or low
-
accuracy waypoints,
including ability to stay in travel lane through road
-
following by
sensing
berms

or road edges, or by any other sensor
-
based
technique.


What did we get from DARPA?


Route Network Definition File (RNDF)


Defines where the roads start and stop


Defines which road connects to which road


Declares the road width, number of lanes,
directionality.


Provides sparse waypoints along the road.


Defines some waypoints as checkpoints


Mission Definition File (MDF)


Declared the speed limits of the roads


Defined the order in which checkpoints are to be
visited


RNDF_name

Team_Case_Site_Visit_RNDF


num_segments

3


num_zones

0


format_version

1.0


creation_date

3/15/2007


segment

1


num_lanes

2


segment_name

Loop_Rd


lane

1.1


num_waypoints

20


lane_width

15


left_boundary

double_yellow


checkpoint

1.1.3

8

checkpoint

1.1.8

9

checkpoint

1.1.13

10

checkpoint

1.1.18

11

stop

1.1.20


exit

1.1.20

1.1.1

exit

1.1.20

2.2.1

exit

1.1.20

3.1.1

1.1.1

41.051285

-
81.470392

1.1.2

41.051196

-
81.470214

1.1.3

41.051106

-
81.470036

1.1.4

41.051018

-
81.469858

1.1.5

41.050929

-
81.469680

1.1.6

41.050832

-
81.469653

1.1.7

41.050753

-
81.469718

1.1.8

41.050675

-
81.469785

1.1.9

41.050595

-
81.469851

1.1.10

41.050517

-
81.469917

1.1.11

41.050497

-
81.470045

1.1.12

41.050585

-
81.470222

1.1.13

41.050673

-
81.470399

1.1.14

41.050761

-
81.470576

1.1.15

41.050849

-
81.470753

1.1.16

41.050947

-
81.470782

1.1.17

41.051027

-
81.470716

1.1.18

41.051105

-
81.470651

1.1.19

41.051184

-
81.470584

1.1.20

41.051263

-
81.470518

end_lane



lane

1.2


num_waypoints

20


lane_width

15


left_boundary

double_yellow


checkpoint

1.2.3

2

checkpoint

1.2.8

3

checkpoint

1.2.13

4

checkpoint

1.2.18

5

stop

1.2.20


exit

1.2.20

1.2.1

exit

1.2.20

2.2.1

exit

1.2.20

3.1.1

1.2.1

41.051286

-
81.470564

1.2.2

41.051207

-
81.470629

1.2.3

41.051128

-
81.470695

1.2.4

41.051049

-
81.470761

1.2.5

41.050969

-
81.470828

1.2.6

41.050816

-
81.470783

1.2.7

41.050728

-
81.470606

1.2.8

41.050640

-
81.470428

1.2.9

41.050552

-
81.470250

1.2.10

41.050463

-
81.470072

1.2.11

41.050495

-
81.469872

1.2.12

41.050574

-
81.469806

1.2.13

41.050653

-
81.469740

1.2.14

41.050732

-
81.469674

1.2.15

41.050810

-
81.469607

1.2.16

41.050963

-
81.469652

1.2.17

41.051052

-
81.469829

1.2.18

41.051140

-
81.470007

1.2.19

41.051230

-
81.470186

1.2.20

41.051319

-
81.470363

end_lane



end_segment



segment

2


num_lanes

2


segment_name

South_Stub_Rd


lane

2.1


num_waypoints

3


lane_width

15


left_boundary

double_yellow


checkpoint

2.1.2

7

stop

2.1.3


exit

2.1.3

1.1.1

exit

2.1.3

1.2.1

exit

2.1.3

3.1.1

2.1.1

41.051581

-
81.470989

2.1.2

41.051482

-
81.470789

2.1.3

41.051384

-
81.470590

end_lane



lane

2.2


num_waypoints

3


lane_width

15


left_boundary

double_yellow


checkpoint

2.2.2

6

exit

2.2.3

2.1.1

2.2.1

41.051418

-
81.470563

2.2.2

41.051516

-
81.470762

2.2.3

41.051615

-
81.470961

end_lane



end_segment



segment

3


num_lanes

2


segment_name

West_Stub_Rd


lane

3.1


num_waypoints

3


lane_width

15


left_boundary

double_yellow


checkpoint

3.1.2

12

exit

3.1.3

3.2.1

3.1.1

41.051416

-
81.470392

3.1.2

41.051569

-
81.470265

3.1.3

41.051721

-
81.470138

end_lane



lane

3.2


num_waypoints

3


lane_width

15


left_boundary

double_yellow


checkpoint

3.2.2

1

stop

3.2.3


exit

3.2.3

1.1.1

exit

3.2.3

1.2.1

exit

3.2.3

2.2.1

3.2.1

41.051744

-
81.470183

3.2.2

41.051591

-
81.470310

3.2.3

41.051439

-
81.470437

end_lane



end_segment



end_file



MDF_name

NE_Stub_To_Outside_To_Inside

RNDF

Team_Case_Site_Visit_RNDF

format_version

1.0

creation_date

09/12/2007

checkpoints


num_checkpoints

3

1

2

9

end_checkpoints




speed_limits




num_speed_limits

3

1

0

9

2

0

9

3

0

9

end_speed_limits



end_file




How Does DEXTER do his thing?


Biologically inspired


Sensory fusion


Behaviors and Moods


Physical State estimation


LIDAR, GPS, IMU, Cameras, Wheel Encoders, RADAR


Everything programmed in LabVIEW


Easy to learn


Easy to write multithreaded applications


Easy to communicate between parts of architecture


How Did DEXTER do?


We had a couple of bugs that we quickly fixed when we
found them.


We ran into a wall twice.


At one point a sensor got aimed wrong and we thought
the road was an obstacle.



We made it to the last 19 teams.


Only 11 teams qualified


We held our own with the big fish and had a good
time.

videos