marcin pilat

Project Archive

 

This page is dedicated to projects that I've done in my courses and my research.

Any programs and documents downloaded from this page are for personal/academia use only and are copyright of the respected authors. Commercial use is prohibited. Please contact me for any related enquiries. Use at own risk. Some programs and source code are licensed under the GNU GPL. Some exe files might require the MFC library from Microsoft(R) Corp. which is not provided.

Morphids

[2007-]

 

For information about the Morphid Academy project please visit the Morphids site.

SwarmArt Exhibitions

[2004]

 

I am a member of the SwarmArt team (swarmart.com) and in 2004, I was involved with two SwarmArt exhibition installations. For these exhibitions, I wrote a totem program that received pictures from a wall-mounted camera and built graphical totems from motion captures in the video. The graphical output was projected onto a wall surface for viewing (and interacting with). The project was managed by Gerald Hushlak, Christian Jacob, and Jeffrey Boyd with the help of Paul Nuytten and Maxwell Sayles.

Exhibition 1: The Other Gallery, The Banff Centre, Banff - August 2004 [picture]

Exhibition 2: The Nickle Arts Museum, University of Calgary - November 2004 [picture]

  • Christian J. Jacob, Gerald Hushlak, Jeffrey E. Boyd, Paul Nuytten, Maxwell Sayles, Marcin Pilat. SwarmArt: Interactive Art from Swarm Intelligence. Leonardo, Vol. 40, No. 3, pages 248-254, 2007. [journal]
  • Wasp-Inspired Construction Algorithms

    [2004]

    download pdf

    Work completed for the Emergent Computing course (CPSC 607) at University of Calgary.

    Abstract: In this paper, we present our research into wasp-inspired construction al gorithms. We were able to implement a typical construction algorithm in a 3D simulation environment and reproduce the results of previous research in the area. Finally, we present case studies that consider the impact of rule changes on the resulting evolved architectures.

  • Marcin L. Pilat. Wasp-Inspired Construction Algorithms. Department of Computer Science, University of Calgary, 2006. (TN 2006-847-40) [report]
  • Simplified models of Dictyostelium discoideum aggregation and slug migration

    [2003]

    download pdf

    Work completed for the Bytes of Life Course (CPSC 605) at Univesity of Calgary.

    Abstract: Biological models of the slime mold Dictyostelium discoideum behaviours are often quite complex. Computer simulations using simplified models provide insight into the biological behaviours but rely on complex mathematics. We studied whether characteristic slime mold behaviours can occur in very simple models that can be easily implemented with any graphical simulation engine. We proposed two excitation based models: neighbour and spherical models, and one simplified cAMP propagation model - the grid model. We succeeded in producing streaming and spiral patterns in aggregation and characteristic thermotaxis in the slug migration life stage of Dictyostelium discoideum.

  • Marcin L. Pilat. Simplified models of Dictyostelium discoideum aggregation and slug migration. Department of Computer Science, University of Calgary, 2006. (TN 2006-844-37) [report]
  • Hierarchical Learning Systems: Robotic Control Using Hierarchical Genetic Programming

    [2003]

    download docs download code

    Masters thesis completed for the M.C.S. degree awarded at Carleton University.

    Abstract: In this thesis, we study the use of hierarchical genetic programming techniques to evolve robotic controllers for a simulated Khepera miniature robot. We study GP chromosome representation methods of linear-genome and tree-based and HGP techniques of Automatically Defined Functions, Module Acquisition, and Adaptive Rep- resentation. We train robotic controllers in the tasks of obstacle avoidance, wall following, and light avoidance. Our results enable us to compare and contrast the five studied representation methods and to provide suggestions for their improvement.

    Information about the simulator is available on the Khepera GP Simulator page .

  • Marcin L. Pilat and Franz Oppacher. Evolution of Khepera Robotic Controllers with Hierarchical Genetic Programming Techniques. In Nadia Nedjah and Luiza de Macedo Mourelle, editors, Evolvable Machines: Theory & Practice, volume 161 of Studies in Fuzziness and Soft Computing, chapter 3, pages 43–71. Springer, Berlin , 2004. [bib] [pdf]
  • Marcin L. Pilat and Franz Oppacher. Robotic Control Using Hierarchical Genetic Programming. In K. Deb et al., editors, Proceedings of the Genetic and Evolutionary Computation Conference – GECCO-2004, Part II, volume 3103 of Lecture Notes in Computer Science, pages 642–653, Springer-Verlag, 2004. [bib] [pdf]
  • Marcin L. Pilat. Hierarchical Learning Systems: Robotic Control Using Hierarchical Genetic Programming.  M.C.S. Thesis, 2003. [bib] [pdf]
  • Using Genetic Algorithms to optimize ACS-TSP

    [2002]

    download pdf

    Work completed for the Swarm Intelligence Course (95.590Y) at Carleton University.

    The paper has been accepted and presented at ANTS'2002. It is published in the conference proceedings as part of the Springer's Lecture Notes in Computer Science (LNCS) series [2463]:

    Abstract: We propose a modification to the ACS-TSP algorithm to improve its performance. This ACSGA-TSP algorithm is a hybrid between ACS-TSP and a genetic algorithm that encodes experimental variables in ants. The algorithm does not yield improved results but offers concepts that can be used to improve the ACS algorithm. We also propose a Meta ACS-TSP algorithm that uses a genetic algorithm to evolve experimental variable values used in ACS-TSP. Through our experimentation we have found that the performance of ACS-TSP can be improved by modifying the variables to suggested values.

  • Marcin L. Pilat and Tony White. (2002) Using Genetic Algorithms to Optimize ACS-TSP. Proceedings of Ant Algorithms: Third International Workshop, ANTS 2002, Brussels, Belgium, September 2002, M. Dorigo et al (Eds.), LNCS 2463, Springer-Verlag, 282-287, 2002. [bib] [pdf]
  • Collective Robotics

    [2002]

    download pdf

    Work completed for the Swarm Intelligence Course (95.590Y) at Carleton University

    Lecture on Collective Robotics.

    Army Ant Model in NetLogo

    [2002]

    download pdf

    Work completed for the Swarm Intelligence Course (95.590Y) at Carleton University.

    A simulation of the Army Ant foraging model in NetLogo. Army Ants in their natural habitats form unique raid foraging patterns. I tried to reproduce those patterns by encoding a model of the foraging in NetLogo. The project was a success since I was able to generate the specific raid patterns of Army Ants. I have also experimented with various model parameters to see the impact they have on the outcome of the simulation.

    Note: NetLogo Beta 10 or later required.

    Theta Graph and q-Sink Spanners

    [2002]

    Joined work with Aaron Lee for the Geometric Networks Course (95.590L) at Carleton University.

    Our project for the Geometric Networks class was to create a visual implementation of theta graphs and q-sink spanners. Those graphs are specific spanner graphs with certain spanner properties. We have used JAVA to create an interactive applet that lets the user create and modify theta graphs. The user can also create q-sink spanners from specific theta graphs. To our knowledge, our work is the only visual implementation of theta graphs and q-sink spanners. Very useful when studying those graphs.

    Note: Shift+Left click marks a node as q.

    Run applet: here

    Euclidean Spanners - Short, Thin, and Lanky

    [2001]

    download pdf

    Work completed for the Computational Geometry Course (95.508) at Carleton University.

    Paper review of:

    • S. Arya, G. Das, D. M. Mount, J. S. Salowe, M. H. Smid. Euclidean Spanners: Short, Thin, and Lanky. In Proceedings of the 27th Annual ACM Symposium on the Theory of Computing, pages 489-498, 1995.

    Robot Motion Planning

    [2001]

    download pdf

    Work completed for the Computation Geometry Course (95.508) at Carleton University.

    For my course project, I chose to do an implementation project based on algorithms presented in Chapter 13: Robot Motion Planning of the Computational Geometry textbook[1]. The reason behind the project was to enable me to study a robot path planning algorithm in detail. s

    • [1] M. de Berg, M. van Kreveld, M. Overmars, O. Schwarzkopf. Computational Geometry - Algorithms and Applications, 2nd edition. Springer, 2000.

    Hats: Separating Roles of an Object in OOD

    [2001]

    download pdf

    Work completed for the Design Patterns Course (95.540) at Carleton University.

    The Hats pattern tries to improve the structure and efficiency of an Object Oriented System Design. It separates the roles that an object has (”Hats”) from the internal state of the object (”Data Object”). The Hats implement the behavior of their roles and use the Data Object to get its internal state. Hats can be used by different Clients in the application that do not need to know or care about the other roles the object has. This abstraction provides an effective separation of roles, both structurally and behavior-wise.

    Network Software License Manager and Application Security Library

    [2000]

    download pdf

    Work completed for the Computer Security and Cryptography Course (95.413) at Carleton University.

    With increasingly more people using technology, software piracy is a major concern of software companies. Companies need to protect their products against unwanted use. One possible protection mechanism is a license server that distributes licenses to clients running on a local network. Such a system would need to be protected against security breaches. Using encryption algorithms, hash functions, and protected code, a security system can be built that is immune to regular software attacks.

    VRML Memory Game

    [2000]

    download pdf

    Work completed for the Multimedia Systems Course (95.410) at Carleton University.

    Sample VRML memory game.

    Decision Tree Induction (ID3, C4.5)

    [2000]

    download pdf

    Work completed for the Artificial Intelligence Course (95.407) at Carleton University.

    The program demonstrates the use of Decision Tree Induction (DTI) algorithm to create a decision tree for a given set of raw data. The program can also be used to test a percentage of the given data (called the testing set) against the tree produced by the rest of the data (called the training set) to come up with percentage accuracy for data classification. Another feature of the program is a time counter that times each decision tree creation and can be used to measure the dependency of input data size on the creation time.

    GA Knapsack Solver

    [2000]

    download pdf

    Work completed for the Artificial Intelligence Course (95.407) at Carleton University.

    The GA Knapsack Solver demonstrates the operation of a genetic algorithm by trying to solve the well known 0-1 knapsack problem.

    Contact Marcin Pilat | ©2013 Marcin Pilat