The 2011 Simulated Car Racing Championship @ Evo*-2011

The 2011 Simulated Car Racing
Championship @ Evo*-2011
Daniele Loiacono, Luigi Cardamone,
Martin V. Butz, and Pier Luca Lanzi

2011 Simulated Car Racing Championship @ EVO*-2011

2

2011 Simulated Car Racing Championship
9 races during 3 conferences

Develop a driver for TORCS
(hand-coded, learned, evolved, …)

Drivers will be awarded based on their score
in each conference competition

At the end, the team with highest overall score
wins the championship


Roadmap of the 2011 Championship

1. EVO*-2011, Torino (Italy)
April 27-29, 2011 (Now!)

2. ACM GECCO-2011, Dublin (Ireland)
July 12-16, 2011

3. IEEE CIG-2011, Seoul (South Korea)
August 31 September 3, 2011


Motivations

 Proposing a relevant game-based competition
more representative of commercial games AI
more similar to a real-world problem

 Proposing a funny and exciting competition
you can see and play with the entries of this competition
human players can interact with AI
a lot of programmed AI available for comparison

 Proposing a challenging competition
computationally expensive
real-time
on-line learning
noisy sensors


The Open Racing Car Simulator

 TORCS is a state of the art open source simulator written in C++

 Main features
Sophisticated dynamics
Provided with several
cars, tracks, and
controllers
Active community of
users and developers
Easy to develop your
own controller

 OS Support
Linux: binaries and building from sources
Windows: binaries and “limited” bulding from sources support
OSX: legacy binaries and no building from sources support 


The Open Racing Car Simulator
& the Competition Software

TORCS TORCS
PATCH

BOT BOT BOT SBOT SBOT SBOT

 The competition server
UDP UDP UDP
 Separates the bots from TORCS
 Build a well-defined sensor model
 Works in real-time BOT BOT BOT


Sensors and actuators

 Rangefinders for edges on the track and opponents (with noise)
 Speed, RPM, fuel, damage, angle with track, distance race, position
on track, etc.

 Six effectors: steering wheel [-1,+1], gas pedal [0, +1], brake
pedal [0,+1], gearbox {-1,0,1,2,3,4,5,6}, clutch [0,+1], focus
direction

12

What is the structure of a race?

Three stages: warm up, qualifiers, actual race

During warm-up, each driver can explore the
track and learn something useful

During qualifiers, each driver races alone against
the clock (the best 8 drivers move to the race)

During the race all the drivers race together


The competitors

 Four entries in the first leg:
Mr. Racer, TU Dortmund
Mariscal & Fernández, Málaga University
Ready2Win, Slovak University of Technology - FIIT
Powaah, Blekinge Institute of Technology

 Three more entries from the 2009 championship:
AUTOPIA, Madrid and Granada
COBOSTAR (Lönneker & Butz, University of Würzburg)
POLIMI (Cardamone, Politecnico di Milano)


Mr Racer

Jan Quadflieg, Tim Delbruegger and Mike Preuss
TU Dortmund


Mr. Racer 2011

 Mostly described in the Evo* Paper “Driving Faster Than a
Human Player“

 13 Variables learned offline with the CMA-ES

 Warmup used to learn a model
of the track

 Noise Handling:
Low pass filter on sensor values
2 Regression polynoms are fitted to each side
Resample the polygons


Mr. Racer 2011 - Opponent Handling

 Noise filtering with a lowpass filter

 Observer module interprets filtered sensor readings

 This leads to a recommended speed and an overtaking line
(created on the fly!)

 Planning module incorporates recommended target speed
and overtaking line into the plan

 Lots of nummerical parameters not yet tuned:
left for Gecco Competition


Mariscal & Fernández

 General approach: expert systems improved by multi-
objectives evolutionary computation techniques

 Steering: based on a simple rule follow the track sensor with
biggest distance.


Mariscal & Fernández - Throttle/Brake

 Implemented via a parameterized expert system whose
parameters were optimized via a Multi-Objective
Evolutionary Algorithm (NSGA-II)

 Two objectives:
Maximize distance.
Minimize damage.

 Linear function (Ci are parameters to optimize):
C1 (Speed) + C2 (Distance to the next turn) + C3 (Turn degrees)

 We obtain three sets of parameters
(1) To run fast
(i.e., controllers) to be used under (2) To run safe
specific conditions of the race (3) To run on dirty
tracks.


Mariscal & Fernández - Overtaking

 Expert system with four groups of rules:

 Other strategies are implemented (via expert
systems) such as collision avoidance and
slipstream overtakings.


Ready2Win

 Modular architecture:
Driving module:
• The speed is computed using the maximum braking distance
Overtaking module
Recovery module
ABS module

 Track learning during the warm-up:
First lap, driven slow, to identify turns (start, end, entry
position, curvature) and learn the track model
Other laps, speed adaptation:
• Increasing according to lateral speed
• Reducing when the car goes off-road


Powaah

Tim Uusitalo
Blekinge Institute of Technology
Supervisor: Stefan Johanson

Powaah

 Based on Artificial Potential Field

 Every sensor in the span (-40, +40) has a charge

 The charge is determined by how far the sensor measures
the track

 Coulomb’s law is used to calculate potentials in the field
 Pot = (C1 * C2 ) / d2

 The sensor with the highest potential is used for computing
how much steering is needed


AUTOPIA

Industrial Computer Science Department.
Centro de Automática y Robótica
Consejo Superior de Investigaciones Científicas
Madrid, Spain
Contact:E. Onieva (enrique.onieva@car.upm-csic.es)

AUTOPIA

 Fuzzy Architecture based on three basic modules for gear,
steering and speed control
optimized with a genetic algorithm

 Learning in the warm-up stage:
Maintain a vector with as many real values as tracklength
in meters.
Vector initialized to 1.0
If the vehicle goes out of the track or suffers damage
then multiply vector positions from 250 meters before the
current position by 0.95.

 During the race the vector is multiplied by F to make the
driver more cautious in function of the damage:
F=1-0.02*round(damage/1000)


COBOSTAR

Thies Lönneker and Martin V. Butz
University of Würzburg

http://www.coboslab.psychologie.uni-wuerzburg.de

COBOSTAR

 One of the best controller of the 2009 Championship

 Parameterized controller optimized with CMA-ES

 Dynamically saves crash points


CIG-2008 Champ

Luigi Cardamone
Politecnico di Milano


Luigi Cardamone

 Winner of the CIG-2008 Car Simulated Competition

 The controller is based on a neural network evolved with
NEAT

 Very simple policy for overtaking


Scoring process: Warm-up & Qualifying

 Scoring process involves three tracks:
OldPeak
Polimi
HillRoad

 The tracks are not distributed with TORCS:
Generated applying evolutionary computation
The competitors cannot know the tracks

 Each controller raced for 100000 game ticks in the warm-up
stage and then its performance is computed in the qualifying
stage as the distance covered within 10000 game ticks


Qualifying: OldPeak

Autopia 8202,92

Cardamone 7372,23

Cobostar 6799,89

Mariscal&Fernan
5062,1
dez

Mr. Racer 8159,67

Powaah 4606,09

Ready2Win 7441,21

0 1000 2000 3000 4000 5000 6000 7000 8000 9000


Qualifying: Polimi

Autopia 8307,57

Cardamone 96,1469

Cobostar 6694,85

Mariscal&Fern
7775,64
andez

Mr. Racer 8570,81

Powaah 3891,01

Ready2Win 6343,6

0 1000 2000 3000 4000 5000 6000 7000 8000 9000


Qualifying: HillRoad

Autopia 11088,4

Cardamone 8937,33

Cobostar 10006,7

Mariscal&Fern
10547,8
andez

Mr. Racer 10250,3

Powaah 10075,3

Ready2Win 7528,29

0 2000 4000 6000 8000 10000 12000


Qualifying summary

Driver OldPeak Polimi HillRoad Total

Autopia 10 8 10 28

Mr. Racer 8 10 6 24

Mariscal&Fernandez 3 6 8 17

Cobostar 4 5 4 13

Ready2Win 6 4 2 12

Cardamone 5 2 3 10

Powaah 2 3 5 10


How much does noise affect the
performance?


HillRoad with and without noise

Autopia

Cardamone

Cobostar

Mariscal&Fernandez

Mr. Racer

Powaah

Ready2Win

0 2000 4000 6000 8000 10000 12000

NO-NOISY NOISY


46

Three Tracks

For each track we run 7 races
with random starting grids

Each race is scored using the F1 point system
(10 to first, 8 to second, 6 to third, …)

Two points to the controller with lesser damage

Two points for the fastest lap of the race


Race: OldPeak

Competitor Score

Autopia 12

Mr. Racer 8

COBOSTAR 6

Cardamone 5

Ready2Win 5
Mariscal&Fer
nandez 4

Powahh 2


Race: Polimi

Competitor Score

Autopia 10
Mariscal&Fer
nandez 8

Mr. Racer 8

COBOSTAR 7

Cardamone 4

Powahh 3

Ready2Win 3


Race: HillRoad

Competitor Score

Autopia 12
Mariscal&Fer
nandez 8

Mr. Racer 6

Powahh 5

COBOSTAR 4

Cardamone 3

Ready2Win 2


Final Results

Competitor OldPeak Polimi HillRoad Total

Autopia 12 10 12 34

Mr. Racer 8 8 6 22

Mariscal&Fernandez 4 8 8 20

COBOSTAR 6 7 4 17

Cardamone 5 4 3 12

Powaah 2 3 5 10

Ready2Win 5 3 2 10


What about the race results?
Race and Qualifying have similar outcomes
(with the exception of Ready2Win)

AUTOPIA still very competitive

Opponent manegement is crucial on race
start. Very difficult with noise!

Thank you!

Looking forward to see your
controllers @ Gecco-2011

The 2011 Simulated Car Racing Championship @ Evo*-2011

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Andere mochten auch

Andere mochten auch (15)

Ähnlich wie The 2011 Simulated Car Racing Championship @ Evo*-2011

Ähnlich wie The 2011 Simulated Car Racing Championship @ Evo*-2011 (20)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

The 2011 Simulated Car Racing Championship @ Evo*-2011