This paper presents the design and implementation of a
control system for a servo motor used for position control
of aerodynamic control surfaces. The system is based on the
microcontroller STM32F411, which performs real-time control
and monitoring of the actuator. The servo motor is equipped
with a lead screw reducer and drives a control surface
characterized by a hinge moment acting as a load
disturbance. The control algorithm is implemented using
proportional integral (PI) and lag compensators in order to
achieve stable and accurate tracking of the reference
signal. Feedback information about the motor position is
obtained from an incremental encoder. In addition, a
display connected to the microcontroller provides real-time
visualization of the encoder position and the reference
value, enabling monitoring of the system during operation.
The proposed control structure is implemented and tested on
an embedded platform. Experimental results demonstrate that
the designed controller achieves satisfactory tracking
performance and stable system behavior despite the presence
of mechanical load and lead screw dynamics.

The most used commutation scheme for electronic commutation
of BLDC motors is trapezoidal (six-step) commutation. It is
simple to implement, and is mostly used for application
that require speed control. However, trapezoidal
commutation can still be used with application that require
other types of control such as position or current. This
paper will address one method of current control of BLDC
motor.

This paper presents the development and experimental
validation of an autonomous navigation system for the
Arduino Alvik mobile robot, designed to find an exit in a
maze
without prior knowledge of its structure. The system
integrates additional sensors, odometry-based localization,
and a proportional-derivative controller for stable
corridor tracking. Maze solving is achieved using a
left-hand wall-following algorithm. Experimental results
demonstrate reliable obstacle detection and stable motion,
confirming the effectiveness of the proposed approach for
autonomous mobile robotics applications.

In this paper, modified PI current controller structures
and a method for tuning the parameters of an adaptive
controller in a drive system with a Surface Permanent
Magnet Synchronous Motor with low moment of inertia are
proposed, taking into account both electrical and
mechanical dynamics. The proposed controller structures
enable the cancellation of cross-coupling between the axes
and do not rely on the calculated speed obtained from
encoder outputs, which may slow down the operation of the
current control loop. The controller performance is
evaluated through computer simulations in both continuous
and discrete domains, as well as through experimental
validation.

This paper presents a deployment-oriented transformation of
Model Predictive Control (MPC) into its explicit form for
cleanroom pressure cascade systems. Starting from an
experimentally validated multivariable MPC controller, an
equivalent explicit MPC (eMPC) formulation is obtained
through offline multi-parametric optimization. In addition
to the explicit-controller derivation, the paper
investigates the relationship between controller
complexity, execution time, and closed-loop performance
through systematic simplification of the explicit solution.
Runtime measurements and performance-degradation analysis
are used to evaluate different explicit-controller
complexities. The results show that the number of regions
can be significantly reduced without measurable degradation
of closed-loop performance according to the proposed
performance degradation index (PDI). At the same time,
measured execution times demonstrate approximately two
orders-of-magnitude runtime speedup compared to
conventional MPC. The paper also discusses architectural
aspects of integrating eMPC into industrial DCS/PLC and
Industrial IoT (IIoT) control environments. The obtained
results indicate that eMPC enables deterministic and
computationally efficient execution suitable for industrial
control environments.

The presence of outliers in measurements can significantly
degrade the performance of standard estimation methods.
Classical approaches, such as Maximum Likelihood (ML),
which rely on the assumption of a correctly specified noise
distribution, become highly sensitive when the actual
distribution deviates from the assumed one. In such
situations, M-estimators exhibit improved performance due
to their robustness to outliers. Although the Huber
contamination model has long been established as a
benchmark in robust estimation, determining the
contamination probability remains a challenging problem.
This paper considers a robust adaptive approach to
estimating the contamination probability based on weighting
coefficients and compares it with classical methods in
scenarios with Gaussian and non-Gaussian noise distribution.

The calorific value of coal is a key parameter in assessing
the production capacity of a thermal power unit. Its
reliable estimation enables optimal tuning of the
combustion process control system. Variations in fuel
quality significantly affect combustion stability, load
distribution among mills, as well as the overall system
efficiency, with their impact reflected through changes in
thermodynamic and electrical quantities. This paper
considers the problem of feature selection and model design
for the estimation of coal calorific value, based on
available process measurements and laboratory-analyzed
samples. Particular focus is placed on the identification
of appropriate physical variables, as well as on
determining the time interval in which they become
representative of the observed sample.

Novel view synthesis has advanced significantly in recent
years, with two main approaches: NeRF-based methods and 3D
Gaussian Splatting. Although both achieve high-quality
results, their performance is yet to be studied and
improved under conditions of degraded inputs. In this
paper, we compare representative methods from both groups.
We degrade inputs by introducing motion blur, a common
effect in real-world image capture, and analyze how it
impacts reconstruction quality. We also analyze how
introducing camera pose errors affects the performance. The
results show how robust each method is and highlight their
strengths and weaknesses when dealing with mentioned input
degradations.

The calculation of the encounter of a projectile and a
target in the air is a common topic in the modern defense
industry. Many methods are used for this purpose, and the
most common are methods related to differential equations
of motion known from the theory of external ballistics,
taking into account the influence of wind, atmosphere, and
thrust from the fuel used to propel the projectile. The
paper presents an example of the calculation of the
prediction of the encounter of a projectile and a target in
the cannon firing zone in an automatic fire control system
based on the integration of differential equations of
motion Ode 45, by the Runge-Kuta method. The obtained
predicted flight time to the target is used to determine
the moment of firing the projectile from the cannon as well
as the azimuth and elevation angles used to direct the
cannon towards the target. Targets at a distance of 3000 m
moving at a uniform speed as well as targets with
acceleration and movement in maneuver were tested. The
simulation results showed good results and a small error in
the projectile passing the target.

Procena kalorijske vrednosti uglja je važan zadatak u
postrojenjima kao što su termoelektrane. Tačnu vrednost je
moguće dobiti jedino u laboratorijskim uslovima, dok se u
realnom vremenu ona može proceniti korišćenjem
odgovarajućih matematičkih modela. Međutim, prisustvo
nepouzdanih merenja može značajno uticati na procenu
parametara modela i tako uticati na smanjenje sposobnosti
generalizacije samog modela. Cilj ovog rada je detekcija
takvih merenja a zatim i ispitivanje njihovog uticaja na
regresioni model za procenu kalorijske vrednosti uglja.
Dobijeni rezultati pokazuju da pristupi iterativne
korekcije modela i usrednjavanja modela predstavljaju dobar
kompromis između tačnosti i stabilnosti parametara čime se
smanjuje osetljivost konačnog modela na izbor i obradu
nepouzdanih merenja.