The aero-pendulum is a non-linear system used broadly to develop and test new controller strategies. This paper presents a new methodology for an adaptive PID fuzzy-based tracking controller using a Hunting Search (HuS) algorithm. The HuS algorithm computes the parameters of the membership functions of the fuzzification stage. As a novelty, the algorithm guarantees the overlap of the membership functions to ensure that all the functions are interconnected, generating new hunters to search for better solutions in the overlapping area. For the defuzzification stage, the HuS algorithm sets the singletons in optimal positions to evaluate the controller response using the centroid method. To probe the robustness of the methodology, the PID fuzzy controller algorithm is implemented in an embedded system to track the angular position of an aero-pendulum test bench. The results show that the adaptive PID fuzzy controller proposed presents root mean square error values of 0.42, 0.40, and 0.49 for 80, 90, and 100 degrees, respectively.

Two-wheeled balancing robots are commonly modeled as inverted pendulum systems, which are inherently unstable and present significant challenges for control due to their nonlinear and underactuated dynamics. One of the main problems in their control is achieving stability and precise position tracking under real-world disturbances and dynamic uncertainties. This research addresses these challenges by proposing a methodology to optimize a linear quadratic regulator (LQR) controller using metaheuristic algorithms. Specifically, the parameters Q and R of the LQR are tuned using three optimization strategies: artificial bee colony (ABC), differential evolution (DE), and global-best harmony search (GHS). The performance of each algorithm is evaluated through a comparative analysis, focusing on control stability and dynamic responsiveness. Experimental results demonstrate that the controller optimized by differential evolution achieves the lowest mean square error (0.4910 degrees), outperforming ABC (0.5187 degrees) and GHS (0.5195 degrees). These findings underscore the effectiveness of metaheuristic algorithms in enhancing control performance for dynamically unstable systems, facilitating their implementation in real-world autonomous control applications.

The optimization of chaotic systems has been performed by considering dynamical characteristics of the mathematical models. The proposed work shows the application of genetic algorithms (GAs) to optimize the chaotic behavior of three well-known systems, namely: Lorenz, Chen and Lü. The parameters of the chaotic systems are varied in a specific range of values considered as the search space, and the evaluation of the mathematical model is performed by applying the Forward Euler method. The contribution presented herein is that the chaotic behavior is evaluated by counting the orbits in an attractor and the sparsity of them. In addition, the chaotic behavior is guaranteed by evaluating the Fourier spectrum of the time series. The solutions provided by the GA, are then implemented on a field-programmable gate array (FPGA) to verify the experimental generation of chaotic attractors. Finally, two optimized chaotic systems are synchronized and used to encrypt an image, thus confirming the appropriateness of optimizing the chaotic behavior by orbit counting and Fourier spectrum analysis.

Two-wheeled self-balancing robots (TWSBRs) are difficult to control because they are nonlinear, unstable, and underactuated, particularly when balance, velocity regulation, and line tracking must be achieved simultaneously. This paper proposes a hybrid parallel control architecture for a line-following TWSBR operating on straight segments, ${90}^{\circ }$ curves, and a ${15}^{\circ }$ slope. Balance stabilization is handled by a classical PD loop, while traslational velocity is regulated by an adaptive fuzzy PI controller, and line following is performed with an adaptive fuzzy PD controller. The fuzzy modules adjust the effective gains based on tracking errors, thereby improving robustness to disturbances, sensor noise, and changes in operating conditions. The complete strategy is implemented on a low-cost PIC18F4550 microcontroller. Experiments show that the fuzzy line-following controller reduces the orientation tracking error compared with a conventional controller. At $0.10{\displaystyle \frac{\mathrm{m}}{\mathrm{s}}}$, RMSE decreases from $0.042\mathrm{rad}$ to $0.038\mathrm{rad}$, and at $0.175{\displaystyle \frac{\mathrm{m}}{\mathrm{s}}}$, it decreases from $0.083\mathrm{rad}$ to $0.066\mathrm{rad}$. The fuzzy approach also improves IAE ($1.317{\displaystyle \frac{\mathrm{rad}}{\mathrm{s}}}$ to $1.185{\displaystyle \frac{\mathrm{rad}}{\mathrm{s}}}$) and ISE ($0.242{\displaystyle \frac{{\mathrm{rad}}^2}{\mathrm{s}}}$ to $0.153{\displaystyle \frac{{\mathrm{rad}}^2}{\mathrm{s}}}$) at $0.175{\displaystyle \frac{\mathrm{m}}{\mathrm{s}}}$, while maintaining similar maximum error ($0.299\mathrm{rad}$ to $0.261\mathrm{rad}$). Overall, the proposed hybrid scheme achieves better adaptability without retuning. These results support real-time deployment on resource-limited platforms.

Two-wheeled self-balancing robots exhibit nonlinear and inherently unstable dynamics due to their inverted-pendulum structure, making control design challenging under terrain variations and external disturbances. This paper proposes a hybrid master–slave fuzzy cascade controller with an additional wheel-synchronization loop to improve tracking performance and robustness. The architecture combines a master velocity PI loop with fuzzy-tuned integral action and a slave balance PD loop with fuzzy proportional control, while a differential synchronization mechanism compensates for motor mismatches without affecting the global balance dynamics. Local stability is analyzed through linearization and equivalent gain approximation within a sector-bounded framework. Experimental validation was conducted on an ESP32-based TWSBR under flat, uphill, and downhill conditions at reference velocities of $0.15$, $0.20$, and $0.30{\displaystyle \frac{\mathrm{m}}{\mathrm{s}}}$, including payload tests with additional masses of $0.279$ and $0.375\mathrm{kg}$. For each scenario, 30 independent trials were performed to compute the reported metrics. Compared with a conventional PID controller, the proposed strategy reduced the flat-terrain velocity RMSE from $0.0108$ to $0.0057{\displaystyle \frac{\mathrm{m}}{\mathrm{s}}}$, while also improving angular stabilization and robustness under slope and payload disturbances.

Learning-based approaches have significantly advanced the capabilities of Simultaneous Localization and Mapping (SLAM) systems, particularly in challenging environments characterized by noise, dynamic objects, and perceptual ambiguity. However, the literature remains highly heterogeneous in terms of sensing modalities, datasets, evaluation protocols, and reporting practices, making systematic comparison difficult. This paper presents a taxonomy-driven review of learning-based SLAM approaches, with particular emphasis on LiDAR-based systems in mobile robotics, and introduces a functional taxonomy that categorizes methods according to the role of learning within the SLAM architecture: (i) learning-enhanced front-end SLAM (T1), (ii) learning-enhanced back-end SLAM (T2), and (iii) learning-centric SLAM systems (T3). Representative studies were analyzed with respect to performance characteristics, robustness, computational requirements, datasets, and deployment-related evidence. The analysis shows that T1 approaches primarily improve local pose estimation and robustness, T2 methods enhance global consistency through learning-based loop closure and relocalization, and T3 approaches explore unified representations, semantic reasoning, and learning-centric autonomy, albeit with greater computational demands and limited deployment evidence. The review further indicates that hybrid approaches combining geometric and learning-based components constitute a prominent trend in the literature, frequently reporting improvements in accuracy and adaptability while maintaining compatibility with established SLAM frameworks. Nevertheless, these observations should be interpreted cautiously, as stronger empirical evidence for hybrid systems may partially reflect their greater technological maturity and broader evaluation history. Finally, the review identifies persistent challenges, including limited cross-domain generalization, high computational requirements, limited deployment-oriented evaluation, and the lack of standardized benchmarking and reporting practices. These findings highlight the need for more reproducible evaluation methodologies, uncertainty-aware learning strategies, and computationally efficient architectures for robust real-world autonomous SLAM.