This paper presents a novel model-based structural optimization approach for the efficient electromechanical development of hexapod robots. First, a hexapod-design-related analysis of both optimization objectives and relevant parameters is conducted based on the derived dynamical model of the robot. A multi-objective optimization goal is proposed, which minimizes energy consumption, unwanted body motion and differences between joint torques. Then, an optimization framework is established, which utilizes a sophisticated strategy to handle the optimization problems characterized by a large set of parameters. As a result, a satisfactory result is efficiently obtained with fewer iterations. The research determines the optimal parameter set for hexapod robots, contributing to significant increases in a robot’s walking range, suppressed robot body vibrations, and both balanced and appropriate motor loads. The modular design of the proposed simulation model also offers flexibility, allowing for the optimization of other electromechanical properties of hexapod robots. The presented research focuses on the mechatronic design of the Szabad(ka)-III hexapod robot and is based on the previously validated Szabad(ka)-II hexapod robot model.

With the advancement of autonomous and electric vehicles, an increasing demand has been observed for the automatic robot-controlled charging of electric vehicles. The idea of developing such charging stations was raised at several research institutions and universities as early as the 2010s, however the appearance of automatic charging stations with higher Technology Readiness Levels (TRL) can only be dated from 2019 onwards. In most of the developed concepts and solutions, a dedicated parking system is required by vehicle drivers, since the operating range of the robots used for charging is limited. In most cases, solutions do not incorporate robots with unique geometries; instead, proven industrial solutions are applied. The robots in these prototypes are typically installed in a fixed position, similar to industrial applications, and are not mobile. The charging of one vehicle is usually performed by one robot. A high-level summary of the developed mechanical system is presented in this project overview. In this research, an automated, robot-controlled electric vehicle charging system was designed, in which vehicles are parked perpendicularly adjacent to each other, and multiple vehicles are charged using a single collaborative robot. The mechanical system was implemented with a robot mounted on an extendable arm attached to a carriage, which is guided in two directions along rails. In this manner, the automatic charging system is positioned precisely at the parking location of the vehicle to be charged.

The deployment of deep learning-based vision systems on edge devices introduces a complex trade-off between computational efficiency and detection accuracy. In this work, we investigate this trade-off in the context of a multi-stage Automatic License Plate Recognition (ALPR) pipeline, evaluated in two heterogeneous edge execution environments: a general-purpose Raspberry Pi 5 single-board computer and the ARTPEC-8 system-on-chip integrated into an Axis smart camera, where neural network inference is accelerated by the on-chip DLPU. All experiments were performed using pre-recorded images loaded from the file system; neither the Axis camera sensor nor a live video stream was used. This study evaluates the impact of model architecture, numerical precision, and input resolution on both inference latency and detection performance. YOLOv5- and YOLOv8-based models were analyzed under multiple quantization schemes (FP32, FP16, dynamic, and INT8), while a cross-platform benchmark was conducted to assess the benefits and limitations of hardware acceleration. The results show that dedicated accelerators provide significant latency reduction at higher resolutions; however, this advantage is accompanied by reduced flexibility and increased sensitivity to quantization effects. In contrast, CPU-based execution enables the use of more recent and quantization-robust model architectures, which can partially compensate for the lack of hardware acceleration when combined with resolution scaling. Furthermore, the analysis hig ights the importance of hybrid-resolution processing in multi-stage pipelines, where different stages can operate at different input resolutions to balance accuracy and performance. The findings demonstrate that optimal system design requires a joint consideration of hardware characteristics, model architecture, and quantization strategy, rather than relying on a single optimization dimension. The presented results provide practical insights for the design of efficient and robust edge-based ALPR systems, with direct implications for real-world industrial deployments.