Design, Implementation, & Impact Jeffrey Santos Hofstra University May, 2010

Introduction  Robocode  Presentation Goals  Presentation Organization

Project Specifications Objectives 1. To design a robot that will effectively destroy CrazyBot 2. To design a tracking algorithm that will accurately predict CrazyBot’s movements based on observations 3. To design a shooting algorithm that will accurately predict where CrazyBot will be when a bullet passes through a potential area of contact

Project Specifications Functional Requirements 1. The robot program should compile and be loadable by the Robocode program. 2. The robot should handle the following events: a) When the radar detects the presence of an enemy robot. b) When the radar no longer detects the presence of an enemy robot. c) When the robot hits a wall or enemy robot. d) When the robot is hit by a bullet

Project Specifications Non-Functional Requirements 1. In a one-to-one battle against CrazyBot, the hit-ratio should be at least 75%. 2. In a one-to-one battle against CrazyBot, the energy level of the robot should be at least 90% when the battle is completed. 3. In a group of ten battles against CrazyBot, the average time to victory should be less than 1200 turns. 4. In a group of ten battles against CrazyBot, the robot should be victorious 100% of the time.

The Algorithms  Tracking  Movement  Targeting Unwritten Rule: Stay Simple!

The Algorithms:Tracking  Keep a detected robot in the radar arc at all times.  Generous radar arc allows for very simple design.  Delegated to targeting algorithm.

The Algorithms:Movement 1. Calculate enemy energy changes. 2. Calculate distance between CrazyDestroyer and target enemy robot. 3. If the energy change is between 0.0 and 3.0, the enemy likely fired on CrazyDestroyer, move at an angle orthogonal to the bearing to the enemy robot a distance of 50 pixels. 4. If the enemy is greater than 300 pixels away from CrazyDestroyer, move towards it.

The Algorithms: Targeting Finding a Circle

The Algorithms: Targeting A “Perfect” Algorithm Mathematically “Perfect” Too complex

The Algorithms: Targeting A Discrete Solution Method of Exhaustion Discrete Sampled Points

The Design

The Design *Data

The Design Circular Path

The Design DataControlSystem

The Design TrackingControlSystem

The Design CrazyDestroyer

Implementation & Testing  Robocode API  Sun Microsystems’ Netbeans 6.8

Implementation & Testing Design Changes  The Geometer Class  Changes to DataControlSystem  BattleData.calcEnemyPosition()

Implementation & Testing Major Bugs Encountered  Departure from a normal relative angle  Data storage and retrieval

Implementation & Testing Unit Testing  CircularPath.calcPosition  Geometer.getAngle  Geometer.calcCircumcenter  Note on Division By Zero

Implementation & Testing Unit Testing: CircularPath.calcPosition PointHeadingVelocityTimeExpected (0,0)111(0.8414, ) (0,0)11( , ) (3, 4)π/412(4.412, ) No Exceptions!

Implementation & Testing Unit Testing: Geometer.getAngle VertexABExpected (0,0)(0,3)(3,0)π/2 (0,0)(3,0)(0,3)π/2 (0,0)(0,3)(0,-3)π (0,0) (0,3)NaN/Infinity (0,0)(0,3)( , 3) The domain of arccosine…?

Implementation & Testing Unit Testing: Geometer.calcCircumcenter ABCExpected (0,1)(1,0)(0,-1)(0,0) (0,1)(0,-1)(1,0)NaN/Infinity (0,1)( , )(1,0)( , ) (1,0)(-1,0)(0,1)NaN/Infinity (0,1)(1,0)(-1,0)(0,0) (0,1)(1,2)(2,3)NaN/Infinity (0,1)(1, )(2,3)( , )

Experimental Results  Experimental Variables Battlefield Size Movement Algorithm  Gathering Data The DataControlSystem 100,000 Trials

Experimental Results Battlefield Size Discrepancies

Experimental Results Movement Algorithm Discrepancies  400x400 Battlefield Constant movement Collisions  800x600/1200x1200 Battlefield High success Successful by design  5000x5000 Battlefield

The Pedagogy of Robocode  Robocode: Assessment For Learning  Extending Robocode to meet the full requirements of an Introductory level Computer Science curriculum  Focus on Secondary Education

Conclusions  The Effectiveness of CrazyDestroyer  The Pedagogy of Robocode

References [1]“Alice”, available at retrieved 04/11/2010. [2]“A Model Curriculum for K-12 Computer Science”, CSTA, available at retrieved 04/11/2010. [3]Clark Aldrich. Learning Online with Games, Simulations and Virtual Worlds: Strategies for Online Instruction. Jossey-Bass, [4]“Computer Science A Course Description”, CollegeBoard AP, available at retrieved 04/11/2010. [5]Douglas Fisher and Nancy Frey. Checking for Understanding: Formative Assessment Techniques for Your Classroom. Prentice Hall, [6]Mathew Nelson, "Robocode", available at retrieved 04/11/2010. [7]Robert J. Marzano, Debra J. Pickering and Jane E. Pollock. Classroom Instruction That Works: Research-Based Strategies For Increasing Student Achievement. Prentice Hall, [8]“Scratch”, available at retrieved 04/11/2010.