Car Racing Competition. Goal Learn or design a controller for TORCS that races as fast as possible alone or in the presence of others drivers “Spiritual.

Car Racing Competition

Goal Learn or design a controller for TORCS that

races as fast as possible alone or in the presence of others drivers

“Spiritual successor” of CEC 2007 Competition

Car Racing Competition meets TORCS More representative of real game AI Better interaction with human players Many good programmed controllers

available Challenges

How to make it easy accessible? Not designed for Machine Learning! More similar to a real-world problem

The Open Racing Car Simulator

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 suppo Windows: binaries and “limited” bulding from sources support OSX: legacy binaries and no building from sources support

Competition API To make TORCS more easy to

use we developed an API based on socket (UDP)

Values of sensors and effectors are sent through UDP

3 components Torcs Patch Server Bot (C++) Client API (C++ and Java)

Server BOT

TORCS

Client Controlller

Client Controlller

Patch

UDP

Sensors

Track Sensors

Opponent Sensors

Rangefinders to sense Opponents Edges of the track

Speed Position on track Rotation speed of wheels RPM Angle with track Distance raced Fuel and damage ...

Effectors Basically 4 effectors

Steering wheel [-1,+1] Gas pedal [0, +1] Brake pedal [0,+1] Gearbox {-1,0,1,2,3,4,5,6}

Rule Hand coding controller were allowed Unknown Track Distance raced by each controller within

10000 tics(200 sec)

The Results

Submissions 5 entries have been submitted to the

competition: Matt Simmerson Leonard Kinnaird-Heether – Wayne State University Chin Hiong Tan – National University of Singapore Diego Perez - University Carlos III, Madrid Simon Lucas – University of Essex

3 more controllers have been considered Daniele’s heuristic C++ controller Julian’s heuristic Java controller Luigi Cardamone - Politecnico di Milano

Scoring setup A server with 2 Quad-core Xeon 2.66 GHz,

8GB RAM, running Fedora Core 6

Scoring process Scoring is a two stage process involving three

tracks (unknown to the competitors): RUUDSKOGEN STREET1 D-SPEEDWAY

In the first stage only a controller at once is tested and performance is defined as the distance covered within 10000 game tics

In the second stage all the controllers (except the three not in the competition) race at the same time and performance is defined simply as the arrival order after 3 laps

First Stage: RUUDSKOGEN

0.00 1000.00 2000.00 3000.00 4000.00 5000.00 6000.00 7000.00

Distance Raced

TAN

DIEGO

SIMON

LUIGI

DANIELE

JULIAN

MATT

LEONARD

RUUDSKOGEN

First Stage: STREET1

0.00 1000.00 2000.00 3000.00 4000.00 5000.00 6000.00 7000.00

Distance Raced

JULIAN

DIEGO

TAN

LEONARD

DANIELE

SIMON

LUIGI

MATT

STREET1

First Stage: D-SPEEDWAY

-2000 0 2000 4000 6000 8000 10000 12000 14000 16000 18000

Distance Raced

DIEGO

JULIAN

DANIELE

TAN

MATT

SIMON

LEONARD

LUIGI

D-SPEEDWAY

First Stage: overall results

RUUDS KOGE N S TRE E T1 D-S PE E DWAY TOTALMATT 8 10 5 23L E ONARD 10 4 8 22L UIG I 4 8 10 22S IMON 3 6 6 15DANIE L E 5 5 3 13J UL IAN 6 1 2 9TAN 1 3 4 8DIE GO 2 2 1 5

F1 point system is used to compute the final scores of the controllers on the 3 tracks

Second Stage: Competition League

RUUDSKOGEN STREET1 D-SPEEDWAY TOTALMATT 10 10 6 26

LEONARD 4 8 10 22SIMON 6 6 8 20

TAN 5 5 5 15DIEGO 5.5 4.5 4 14

Also in this stage, F1 point system is used to compute the final score

Second Stage: Full League

RUUDSKOGEN STREET1 D-SPEEDWAY TOTALMATT 10 8 6 24

LEONARD 3 10 8 21LUIGI 6 4 10 20

SIMON 4 6 5 15DANIELE 6 4 3 13

TAN 4 3 4 11JULIAN 2 5 4 11DIEGO 5 3 2 10

Summary of results 3 learned controllers are significantly

better than programmed controllers Some possible overfitting to training

tracks? Poor performance in avoiding and

overtaking opponents Some problems with recovery behavior Still more work to reach the performance

of controllers provided with TORCS but initial results are promising!

TORCS Official Site

http://torcs.sourceforge.net/ Source Code Download

http://torcs.sourceforge.net/index.php?name=Sections&op=viewarticle&artid=3

Robot Tutorial http://www.berniw.org/