Dornberger, Rolf
E-Mail-Adresse
Geburtsdatum
Suchergebnisse
Optimization of multi-robot sumo fight simulation by a genetic algorithm to identify dominant robot capabilities
This paper analyzes the multirobot sumo fight simulation. This simulation is based on a computational model of several sumo fighters, which physically interact while trying to move the opponent out of the arena (lost fight). The problem is optimized using a genetic algorithm (GA), where the capabilities of not only one particular robot but of all robots simultaneously are improved. In this particular problem setup, the problem definition changes depending on the optimization path, because all robots also get better, competing against each other. The influence of different operators of the GA is investigated and compared. This paper raises the questions, which genetically controlled capabilities (e.g. size, speed) are dominant over time and how they can be identified by a sensitivity analysis using a GA. The results shed light on which parameters are dominant. This experiment typically opens up interesting fields of further research, especially about how to address optimization problems, where the optimization process influences the search space and how to eliminate the factor of randomness.
Optimal learning rate and neighborhood radius of Kohonen's self-organizing map for solving the travelling salesman problem
The Travelling Salesman Problem (TSP) is one of the well-studied classic combinatorial optimization problems and proved to be a non-deterministic polynomial-time (NP) hard problem. Kohonen's self-organizing map (SOM) is a type of artificial neural network, which can be applied on the TSP. The purpose of the algorithm is to adapt a special network to a set of unorganized and unlabeled data so that it can be used for clustering and simple classification tasks. In this paper, we study the effect of changing the parameters in the SOM algorithm to solve the TSP. The focus of the parameter investigation lies on the influence of changes in the SOM learning rate and neighborhood radius as well as on the number of iterations in TSP problems with varying number of cities. Thus, the investigation is based on various problem instances as well as on different parameter settings of the SOM, which are compared with each other and discussed. The results are additionally compared with the nature inspired ant colony optimization (ACO) algorithm. As a result, it is proved that with the right parameter setting the SOM generated result is improved and that it is superior to the ACO algorithm.
Problem-Based Learning in Teaching the Module “Optimization for Business Improvement”
Comparison of the behavior of swarm robots with their computer simulations applying target-searching algorithms
This paper investigates the functionality and quality of the implementation of a search and target surrounding swarm robotic algorithm using physical swarm robots named Kilobots. The implementation was developed and tested in the simulator V-REP, then transferred onto the actually running Kilobots: Ten Kilobots were used for the experiment, where one Kilobot acts as the target and nine Kilobots act as the searchers. The algorithm allows the searchers to swarm out to find the target while avoiding collisions with other searchers, to orbit around other searchers, which are closer to the target, and finally to surround the target once it is found. The results of the implementation using the physical Kilobots are compared with the results of two adjusted computer simulations. Differences between the simulations and the real robot implementation are investigated: Discrepancies regarding the locomotion and the communication capabilities are identified and discussed.
Comparison of a real kilobot robot implementation with its computer simulation focussing on target-searching algorithms
This paper presents the functionality and quality of the implementation of a search- and target-surrounding swarm robotic algorithm, which was developed and tested in the simulator V-REP, on actually running Kilobots. Ten Kilobots were used for the experiment where one Kilobot acts as target and nine Kilobots act as searchers. The algorithm allows the searchers to disperse to find the target, to avoid collisions with other searchers, to orbit other searchers, which are closer to the target, and to finally surround the target, once it is found. The results of the implementation using actual, real swarm robots are compared with the results of the computer simulation. Differences between simulation and real robot implementation are discussed. In particular, issues associated with the limitation in the Kilobots’ communication capability and their implications on the algorithm are investigated.