Cursus , Publications , Research , Software Packages , Teaching
During my research activities, I participated in development of various software packages and simulators. In this context, I was either the supervisor of a project or I contributed as the main developer or just as one of the programmers.
In the following list, you find some of the software packages/simulators in which I had some contributions:
Wrapped Operations Research Methods (WORM):
In order to provide a pedagogic tool for the students of Operations Research (OR) at the Technische Universtät Kaiserslautern (TUK), we started a project with the objective of developing an interactive tool. In this project, we created a GUI containing several algorithms that we teach in the OR lectures of the TUK.
The current version of the tool contains the following algorithms: Dijkstra, Floyd's algorithm, Simplex (a compact version), Cutting Planes, and three branch-and-bound algorithms for solving the binary Knapsack Problem.
The tool has several interesting features e.g., a student can create his/her own examples, modify them, save the solutions, etc. Furthermore, the tool is able to interact with the students, i.e., in case of a wrong operation (step) by the student, he/she will receive appropriate warning message(s) in order to perform correct operation(s).
The new version is available since April 2018 on the platform "Online Learning and Training (OLAT)" and has been downloaded and used by +1000 students.
Automated tour planning for driving service of children with disabilities:
In this project, we have developed a web-based platform to assist the employees of the German Red Cross (Deutsches Rotes Kreuz (DRK)) in planning of driving service to pick up disabled students from home and to bring them to their kindergarten. The platform, which we have developed, uses an efficient heuristic to solve the problem. The outputs of the platform include visualized maps and routes as well as numerical information.
This project was developed in response to the need of the German Red Cross at the city of Landstuhl, state Rhineland-Palatinate in Germany.
The package is available from here.
Last-Mile Deliveries by Autonomous Drones/Robots:
I was the supervisor of an internship through which we have developed a platform in order to simulate and visualize delivering parcels through trucks and micro depots (via autonomous drones). The graphical user interface (GUI) incorporates several interesting features, e.g., adjusting parameter settings. In addition, the GUI uses an advanced hybrid algorithm based on tabu search and classical optimization methods.
Parcel Delivery Planning through Mini-TSP Tours:
In this project, we developed a platform that receives the plan of a city or a given district as well as some usual information. Then, the platform will provide route plans for delivering parcels by postmen (or possibly by autonomous robots). The platform constructs a set of realistic vehicle routes, where each route is consists of many shorter tours, i.e., tours based on solving mini traveling salesman problems (min-TSP). The idea of this project comes from the actual practical context of delivering parcels by postmen at the Deutsche Post.
Optimal Selection of Charging Stations for E-Scooters:
In this project that I have supervised, we have developed a web-based platform, as an assistant tool, that allows collecting e-scooters, which are scattered throughout a city, for an optimal assignment to charging stations. The platform permits inserting location of e-scooters, finds the closest charging stations by means of, either a mathematical model, which can be solved by Gurobi (if any license is available), or an integrated heuristic.
Bandit Learning for Portfolio Selection:
In this project that I have supervised, we have developed a GUI that uses a learning approach (based on bandit learning and combinatorial optimization techniques) for finding an optimal selection of financial assets out of a given set of stocks. The GUI offers the possibility of using stored data sets and choosing different values for the parameters. In addition, the GUI can visualize the solution procedure, present the results, and save them as output in different formats.
A Multi-Agent Platform for Solving the Art Gallery Problem (AGP):
I was the supervisor of a project for developing a platform in order to simulate and visualize the solution procedure of the AGP by means of autonomous (robotic) agents. The package is independent of any external library (such as Matlab or CGAL). This can be considered as an advantage, because the simulator becomes portable and easy to use. Several algorithms have been included in the platform and the current objective consists in extending the simulator in order to include a larger variety of solution methods.
Based on the results of this simulator, an article is accepted to be published in the proceedings of "IEEE SSCI 2016".
The package is available from here (for downloading the simulator, please see the bottom of this page. The name of the file is: AGP_MAS.zip).
The Hierarchical Optimistic Optimization Simulator:
I was the supervisor of a project for developing a platform in order to find the optima of nonlinear functions by means of a machine learning method that is called "Hierarchical Optimistic Optimization (HOO)". We used this approach in portfolio optimization and published a paper (see Portfolio Optimization by Means of a $\chi$-Armed Bandit Algorithm). The simulator platform is also quite useful for teaching purposes and visualizing the HOO.
The package is available from here (for downloading the source codes, please see the bottom of this page. The name of the file is: HOOTestWithGui.zip). This zipped file includes also a jar file in the "dist" folder.
The Cake Cutting Problem:
I was the supervisor of a project for developing a Graphical User Interface (GUI) for solving the Cake Cutting Problem (CCP). The current version of the GUI contains four different algorithms for solving the CCP. The GUI has also different functionalities.
The "Cake Cutting Problem" concerns allocation of separable resources among a finite number of people with heterogeneous preferences. The preferences might be measured in different ways; for example, by utility functions. You might find several research papers dealing with different solution methods applied to solve the CCP. However, there was no software-tool using the introduced algorithms. In this GUI we tried to bridge this gap.
Art Gallery Problem and its Variants:
I participated in a project entitled: "Kunst! - Exact Algorithms for Art Gallery Variants", supported by the DFG Priority Program 1307: Algorithm Engineering. The project was supervised by Alexander Kröller at the Technical University of Braunschweig (Germnay). Several people participated in this project and produced many original scientific documents. You can get more information on the website of the project: "Kunst! - Exact Algorithms for Art Gallery Variants".
Figurative Braid Creator:
I was the supervisor of a project for developing a platform for creating figurative braid from a given image. The work inspires from the article "Figurative Tours and Braids", by Robert Bosch and Tom Wexler.
The package is available from here (for downloading the source codes, please see the bottom of this page. The name of the file is: FBP.zip).