Gábor Kusper
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Gábor Kusper has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
- Csaba Biró
Faculty of Informatics, Eszterházy Károly Catholic University, 3300 Eger, Hungary
4 shared publications - Benedek Nagy
Department of Mathematics, Faculty of Arts and Sciences, Eastern Mediterranean University, Famagusta, Turkey
3 shared publications - Erika Ábrahám
RWTH Aachen University Aachen Germany
2 shared publications - Isabela Dramnesc
West University
2 shared publications - Sorin Stratulat
Université de Lorraine
2 shared publications - Tudor Jebelean
Johannes Kepler University
2 shared publications
Latest Publications
- Open Access
Digital-Twin-Oriented Virtual Training Environment for Agricultural Robot Navigation: A Vineyard Rover Case Study
A virtual training environment offers clear advantages for agricultural robotics. It provides a safe setting in which perception, navigation, and control algorithms can be evaluated without risking damage to either the robot or the crop. It also supports efficient data generation: large volumes of training data can be collected under diverse environmental conditions that would be costly, slow, and often season-dependent in real-world deployments. This broader variability improves model adaptability, reduces the risk of overfitting, and leads to more robust operation. In this paper, we argue that digital twin technology should therefore be understood not merely as a passive mirror of a physical robot, but as an active training environment in which multiple sensor-related subprocesses can be developed, tested, validated, and refined jointly. This paper is based on our experiences with digital twin technology used in the development of a vineyard robot, including a self-driving rover, sensor simulation, procedural map generation, and agriculture-specific movement models. Our contribution is threefold: we reinterpret the digital twin as a training space, propose a layered framework for training agricultural robots in virtual environments, and explain why agriculture is a particularly strong use case, given variable field conditions, expensive real-world experimentation, and persistent labor scarcity. To validate this framework, we present the simulation-based evaluation of an autonomous reinforcement learning agent. The agent has been trained entirely in this virtual environment, which successfully navigated to 155 out of 161 target points in a simulated vineyard demonstration environment.
Open AccessDigital-Twin-Oriented Virtual Training Environment for Agricultural Robot Navigation: A Vineyard Rover Case Study
A virtual training environment offers clear advantages for agricultural robotics. It provides a safe setting in which perception, navigation, and control algorithms can be evaluated without risking damage to either the robot or the crop. It also supports efficient data generation: large volumes of training data can be collected under diverse environmental conditions that would be costly, slow, and often season-dependent in real-world deployments. This broader variability improves model adaptability, reduces the risk of overfitting, and leads to more robust operation. In this paper, we argue that digital twin technology should therefore be understood not merely as a passive mirror of a physical robot, but as an active training environment in which multiple sensor-related subprocesses can be developed, tested, validated, and refined jointly. This paper is based on our experiences with digital twin technology used in the development of a vineyard robot, including a self-driving rover, sensor simulation, procedural map generation, and agriculture-specific movement models. Our contribution is threefold: we reinterpret the digital twin as a training space, propose a layered framework for training agricultural robots in virtual environments, and explain why agriculture is a particularly strong use case, given variable field conditions, expensive real-world experimentation, and persistent labor scarcity. To validate this framework, we present the simulation-based evaluation of an autonomous reinforcement learning agent. The agent has been trained entirely in this virtual environment, which successfully navigated to 155 out of 161 target points in a simulated vineyard demonstration environment.
- Open Access
- 16 January 2026
- Published by MDPI AG in Mathematics
- Vol. 14 (2) , 317
Introducing a Resolvable Network-Based SAT Solver Using Monotone CNF–DNF Dualization and Resolution
This paper is a theoretical contribution that introduces a new reasoning framework for SAT solving based on resolvable networks (RNs). RNs provide a graph-based representation of propositional satisfiability in which clauses are interpreted as directed reaches between disjoint subsets of Boolean variables (nodes). Building on this framework, we introduce a novel RN-based SAT solver, called RN-Solver, which replaces local assignment-driven branching by global reasoning over token distributions. Token distributions, interpreted as truth assignments, are generated by monotone CNF–DNF dualization applied to white (all-positive) clauses. New white clauses are derived via resolution along private-pivot chains, and the solver’s progression is governed by a taxonomy of token distributions (black-blocked, terminal, active, resolved, and non-resolved). The main results establish the soundness and completeness of the RN-Solver. Experimentally, the solver performs very well on pigeonhole formulas, where the separation between white and black clauses enables effective global reasoning. In contrast, its current implementation performs poorly on random 3-SAT instances, highlighting both practical limitations and significant opportunities for optimization and theoretical refinement. The presented RN-Solver implementation is a proof-of-concept which validates the underlying theory rather than a state-of-the-art competitive solver. One promising direction is the generalization of strongly connected components from directed graphs to resolvable networks. Finally, the token-based perspective naturally suggests a connection to token-superposition Petri net models.
- 16 January 2026
- Published by MDPI AG in Mathematics
- Vol. 14 (2) , 317
Open AccessIntroducing a Resolvable Network-Based SAT Solver Using Monotone CNF–DNF Dualization and Resolution
This paper is a theoretical contribution that introduces a new reasoning framework for SAT solving based on resolvable networks (RNs). RNs provide a graph-based representation of propositional satisfiability in which clauses are interpreted as directed reaches between disjoint subsets of Boolean variables (nodes). Building on this framework, we introduce a novel RN-based SAT solver, called RN-Solver, which replaces local assignment-driven branching by global reasoning over token distributions. Token distributions, interpreted as truth assignments, are generated by monotone CNF–DNF dualization applied to white (all-positive) clauses. New white clauses are derived via resolution along private-pivot chains, and the solver’s progression is governed by a taxonomy of token distributions (black-blocked, terminal, active, resolved, and non-resolved). The main results establish the soundness and completeness of the RN-Solver. Experimentally, the solver performs very well on pigeonhole formulas, where the separation between white and black clauses enables effective global reasoning. In contrast, its current implementation performs poorly on random 3-SAT instances, highlighting both practical limitations and significant opportunities for optimization and theoretical refinement. The presented RN-Solver implementation is a proof-of-concept which validates the underlying theory rather than a state-of-the-art competitive solver. One promising direction is the generalization of strongly connected components from directed graphs to resolvable networks. Finally, the token-based perspective naturally suggests a connection to token-superposition Petri net models.
- 1 June 2025
- Published by IATED Academy in EDULEARN Proceedings
- Vol. 1, 1221-1228
AIROBO: A EUROPEAN PROJECT FOR SOFTWARE ENGINEERS
- 1 June 2025
- Published by IATED Academy in EDULEARN Proceedings
- Vol. 1, 1221-1228
AIROBO: A EUROPEAN PROJECT FOR SOFTWARE ENGINEERS
- 1 January 2025
- Published by Eszterhazy Karoly Egyetem
- p. 78-89
LLM-based framework to support the construction of valid formal models
The use of large language models (LLMs) in software development is becoming increasingly widespread, despite well-known concerns regarding their reliability. A significant risk arises from relying on poorly understood approximate solutions that may subtly introduce errors into the final system. A key barrier to the adoption of formal modeling – beyond the steep learning curve of formal specification languages – is the additional abstraction layer, which can be as difficult to maintain as the source code itself. This complexity persists even when the formal specification can generate code directly. Another challenge is that, while tools for verifying properties of formal models are well-established, the initial translation of a mental model into a formal one often results in invalid or imprecise representations. We propose a tool which facilitates the validation of formal models generated by LLMs from natural language specifications. The validation process involves two steps: first, the formal model is translated back into natural language using a deterministic, easily verifiable rule-based method; second, the author of the original specification validates this reformulated version. This human-in-the-loop method mitigates the risks associated with LLM black-box generation by enabling explicit semantic verification of the model.
- 1 January 2025
- Published by Eszterhazy Karoly Egyetem
- p. 78-89
LLM-based framework to support the construction of valid formal models
The use of large language models (LLMs) in software development is becoming increasingly widespread, despite well-known concerns regarding their reliability. A significant risk arises from relying on poorly understood approximate solutions that may subtly introduce errors into the final system. A key barrier to the adoption of formal modeling – beyond the steep learning curve of formal specification languages – is the additional abstraction layer, which can be as difficult to maintain as the source code itself. This complexity persists even when the formal specification can generate code directly. Another challenge is that, while tools for verifying properties of formal models are well-established, the initial translation of a mental model into a formal one often results in invalid or imprecise representations. We propose a tool which facilitates the validation of formal models generated by LLMs from natural language specifications. The validation process involves two steps: first, the formal model is translated back into natural language using a deterministic, easily verifiable rule-based method; second, the author of the original specification validates this reformulated version. This human-in-the-loop method mitigates the risks associated with LLM black-box generation by enabling explicit semantic verification of the model.
- Open Access
- 24 August 2023
- Published by Annales Mathematicae et Informaticae - AMI in Annales Mathematicae et Informaticae
- Vol. Accepted m, 92-109
Using extended resolution to represent strongly connected components of directed graphs
- 24 August 2023
- Published by Annales Mathematicae et Informaticae - AMI in Annales Mathematicae et Informaticae
- Vol. Accepted m, 92-109
Open AccessUsing extended resolution to represent strongly connected components of directed graphs
- Open Access
- 1 January 2023
- Published by Neumann Janos Egyetem in Gradus
- Vol. 10 (2)
A Grapler szőlőmetsző robot és a gazda modul tervei
- 1 January 2023
- Published by Neumann Janos Egyetem in Gradus
- Vol. 10 (2)
Open AccessA Grapler szőlőmetsző robot és a gazda modul tervei
- 17 September 2022
- Published by Springer Nature
- p. 287-304
Experiments with Automated Reasoning in the Class
- 17 September 2022
- Published by Springer Nature
- p. 287-304
Experiments with Automated Reasoning in the Class
- Open Access
- 15 October 2021
- Published by MDPI AG in Mathematics
- Vol. 9 (20) , 2597
Resolvable Networks—A Graphical Tool for Representing and Solving SAT
In this paper, we introduce the notion of resolvable networks. A resolvable network is a digraph of subnetworks, where subnetworks may overlap, and the inner structure of subnetworks are not interesting from the viewpoint of the network. There are two special subnetworks, Source and Sink, with the following properties: there is no incoming edge to Source, and there is no outgoing edge from Sink. Any resolvable network can be represented by a satisfiability problem in Boolean logic (shortly, SAT problem), and any SAT problem can be represented by a resolvable network. Because of that, the resolution operation is valid also for resolvable networks. We can use resolution to find out or refine the inner structure of subnetworks. We give also a pessimistic and an optimistic interpretation of subnetworks. In the pessimistic case, we assume that inside a subnetwork, all communication possibilities are represented as part of the resolvable network. In the optimistic case, we assume that each subnetwork is strongly connected. We show that any SAT problem can be visualized using the pessimistic interpretation. We show that transitivity is very limited in the pessimistic interpretation, and in this case, transitivity corresponds to resolution of clauses. In the optimistic interpretation of subnetworks, we have transitivity without any further condition, but not all SAT problems can be represented in this case; however, any such network can be represented as a SAT problem. The newly introduced graphical concept allows to use terminology and tools from directed graphs in the field of SAT and also to give graphical representations of various concepts of satisfiability problems. A resolvable network is also a suitable data structure to study, for example, wireless sensor networks. The visualization power of resolvable networks is demonstrated on some pigeon hole SAT problems. Another important application field could be modeling the communication network of an information bank. Here, a subnetwork represents a dataset of a user which is secured by a proxy. Any communication should be done through the proxy, and this constraint can be checked using our model.
- 15 October 2021
- Published by MDPI AG in Mathematics
- Vol. 9 (20) , 2597
Open AccessResolvable Networks—A Graphical Tool for Representing and Solving SAT
In this paper, we introduce the notion of resolvable networks. A resolvable network is a digraph of subnetworks, where subnetworks may overlap, and the inner structure of subnetworks are not interesting from the viewpoint of the network. There are two special subnetworks, Source and Sink, with the following properties: there is no incoming edge to Source, and there is no outgoing edge from Sink. Any resolvable network can be represented by a satisfiability problem in Boolean logic (shortly, SAT problem), and any SAT problem can be represented by a resolvable network. Because of that, the resolution operation is valid also for resolvable networks. We can use resolution to find out or refine the inner structure of subnetworks. We give also a pessimistic and an optimistic interpretation of subnetworks. In the pessimistic case, we assume that inside a subnetwork, all communication possibilities are represented as part of the resolvable network. In the optimistic case, we assume that each subnetwork is strongly connected. We show that any SAT problem can be visualized using the pessimistic interpretation. We show that transitivity is very limited in the pessimistic interpretation, and in this case, transitivity corresponds to resolution of clauses. In the optimistic interpretation of subnetworks, we have transitivity without any further condition, but not all SAT problems can be represented in this case; however, any such network can be represented as a SAT problem. The newly introduced graphical concept allows to use terminology and tools from directed graphs in the field of SAT and also to give graphical representations of various concepts of satisfiability problems. A resolvable network is also a suitable data structure to study, for example, wireless sensor networks. The visualization power of resolvable networks is demonstrated on some pigeon hole SAT problems. Another important application field could be modeling the communication network of an information bank. Here, a subnetwork represents a dataset of a user which is secured by a proxy. Any communication should be done through the proxy, and this constraint can be checked using our model.
- 19 May 2021
- Published by Annales Mathematicae et Informaticae - AMI in Annales Mathematicae et Informaticae
- Vol. 53, 299-316
Abstractive text summarization for Hungarian
- 19 May 2021
- Published by Annales Mathematicae et Informaticae - AMI in Annales Mathematicae et Informaticae
- Vol. 53, 299-316
Abstractive text summarization for Hungarian
- 19 March 2021
- Published by Annales Mathematicae et Informaticae - AMI in Annales Mathematicae et Informaticae
- Vol. Accepted m, 33-43
Introducing w-Horn and z-Horn: A generalization of Horn and q-Horn formulae
- 19 March 2021
- Published by Annales Mathematicae et Informaticae - AMI in Annales Mathematicae et Informaticae
- Vol. Accepted m, 33-43
Introducing w-Horn and z-Horn: A generalization of Horn and q-Horn formulae