Wyniki wyszukiwania - BazTech

1

Application of artificial neural networks and NEAT algorithm to control a quadcopter

Przychodzki Mateusz, Niewiadomski Artur

Studia Informatica : systems and information technology

|

2023

|

Vol. 2(29)

41--56

EN

We present a framework for building ArtificialNeuralNetworks (ANNs) able to control a quadcopter and perform basic maneuvers, like hover or following waypoints. In this approach, we make use of the Neuroevolution of Augmenting Topologies (NEAT) algorithm which is aimed at creating the network structure and the weights in result of evolutionary computations. In order to evaluate fitness of individuals, we use physics based, realistic simulation engine Gazebo, where each individual controls a drone in a simulated environment. Our approach is aimed at using one of existing, popular protocols used to remotely control drones, and train ANNs able to imitate signals received from a radio controller operated by a human pilot. Thus, contrary to the most of other approaches, our autonomous controller cooperates with standard drone software. Our ultimate goal is to train ANNs able to control a real-world quadcopter and perform advanced tasks autonomously. Not only such ANNs should be able to perform the maneuvers correctly, but they should be small enough to transfer them into a quadcopter’s limited memory. In this paper we report the first stage of our project - a successful development and deployment of the ANNs distributed training framework, and choosing the activation function for further research.

2

Unbounded Model Checking for ATL

Kański Michał, Niewiadomski Artur, Kacprzak Magdalena, Penczek Wojciech, Nabiałek Wojciech

Studia Informatica : systems and information technology

|

2021

|

Vol. 1-2(25)

5--22

EN

In this paper, we deal with verification of multi-agent systems represented as concurrent game structures. To express properties to be verified, we use Alternating-Time Temporal Logic (ATL) formulas. We provide an implementation of symbolic model checking for ATL and preliminary, but encouraging experimental results.

3

Object classification with artificial neural networks : A comparative analysis

Domeradzki Kornel, Niewiadomski Artur

Studia Informatica : systems and information technology

|

2019

|

Vol. 1-2(23)

43--56

EN

Object classification is a problem which has attracted a lot of research attention in recent years. Traditional approach to this problem is built on a shallow trainable architecture that was meant to detect handcrafted features. That approach works poorly and introduces many complications in situations where one is to work with more than a couple types of objects in an image with a large resolution. That is why in the past few years convolutional and residual neural networks have experienced a tremendous rise in popularity. In this paper, we provide a review on topics related to artificial neural networks and a brief overview of our research. Our review begins with a short introduction to the topic of computer vision. Afterwards we cover briefly the concepts of neural networks, convolutional and residual neural networks and their commonly used models. Then we provide a comparative performance analysis of the previously mentioned models in a binary and multi-label classification problem. Finally, multiple conclusions are drawn, which are to serve as guidelines for future computer vision systems implementations.

4

Applying Modern SAT-solvers to Solving Hard Problems

Niewiadomski Artur, Switalski Piotr, Sidoruk Teofil, Penczek Wojciech

Fundamenta Informaticae

|

2019

|

Vol. 165, nr 3-4

321--344

EN

We present nine SAT-solvers and compare their efficiency for several decision and combinatorial problems: three classical NP-complete problems of the graph theory, bounded Post correspondence problem (BPCP), extended string correction problem (ESCP), two popular chess problems, PSPACE-complete verification of UML systems, and the Towers of Hanoi (ToH) of exponential solutions. In addition to several known reductions to SAT for the problems of graph k-colouring, vertex k-cover, Hamiltonian path, and verification of UML systems, we also define new original reductions for the N-queens problem, the knight’s tour problem, and ToH, SCP, and BPCP. Our extensive experimental results allow for drawing quite interesting conclusions on efficiency and applicability of SAT-solvers to different problems: they behave quite efficiently for NP-complete and harder problems but they are by far inferior to tailored algorithms for specific problems of lower complexity.