Tactile texture recognition is a cornerstone of robotic perception, enabling systems to discern and interact with their environments through tactile feedback. In this work, we introduce a comprehensive methodology for texture classification using time-series data acquired from advanced tactile sensors. Our dataset comprises MARG data captured from a variety of distinct textures on uneven surfaces, which poses significant challenges due to variations in material properties and surface irregularities. To address these challenges, we developed a structured pipeline that begins with preprocessing raw sensor signals—implementing noise reduction, normalization, and segmentation—to enhance subsequent feature extraction. While approaches such as 1D-Convolutional Neural Networks (1D-CNNs), Long Short-Term Memory networks (LSTMs), and hybrid CNN-BiLSTM architectures have been recently explored in this domain, our study proposes an innovative informer-based neural network that leverages transformer and attention mechanisms. This architecture is designed to capture both temporal dependencies and spatial patterns inherent in the tactile data more effectively. A rigorous experimental setup employing cross-validation was used to assess the model’s performance and its ability to generalize to unseen surfaces. Experimental results demonstrate that the proposed informer-based model outperforms the existing approaches, with hybrid architectures yielding the highest accuracy. Overall, our contribution advances the field of tactile perception by providing a robust and scalable framework for texture classification, which is pivotal for enhancing the adaptability and precision of robotic systems. Our best model has reached 93.88% accuracy which is the highest accuracy in the literature on uneven surfaces.

The adoption of electric vehicles (EVs) continues to grow, driven by rising fuel prices, environmental concerns, and advancements in battery technology. However, challenges such as limited charging infrastructure and complex route planning still hinder large-scale deployment. This paper addresses the problem of minimizing total travel costs by jointly optimizing route selection, travel time, and charging expenses. An MILP model is introduced for small-scale scenarios, and a scalable heuristic, Minimizing Travel Cost (MTC), is proposed for real-time decisions. MTC integrates real-time traffic and dynamic charging rates, ensuring adaptive, cost-efficient routing without breaching battery safety thresholds. The results show that MTC achieves near-optimal performance with significantly shorter computation time. This work offers a practical and robust solution for intelligent EV routing and charging optimization in real-world transportation systems.

Autonomous vehicles (AVs) are anticipated to play a pivotal role in intelligent transportation systems, particularly in the context of future smart cities. Common performance measures like throughput and latency are not sufficient for capturing the timing and freshness of data in applications like autonomous driving and accident prevention. Therefore, this paper addresses the challenge of minimizing the Age of Information (AoI) in AVassisted vehicular networks. First, the problem is mathematically formulated as linear programming to derive optimal solutions. Recognizing the computational complexity, a scalable heuristic method tailored for large networks is proposed. Additionally, for comparative analysis, the problem is modeled as a Markov decision process and solved using Q-learning, an algorithm of Reinforcement Learning (RL). The numerical results highlight the efficacy of the proposed heuristic method in minimizing the average AoI, considering both computational efficiency and its potential to complement RL algorithms in a hybrid approach.

Modern healthcare should include artificial intelligence (AI) technologies for disease identification and monitoring, particularly for chronic conditions, including heart, diabetes, kidney, liver, and thyroid. According to the World Health Organization (WHO), heart, diabetes, and liver diseases (hepatitis B and C and liver cirrhosis) are leading causes of mortality. The prevalence of thyroid and chronic kidney diseases is also increasing. We conducted a comprehensive review of the available literature to assess the current state of AI advancement in disease diagnosis and identify areas needing further attention. Machine learning (ML), deep learning (DL), and ensemble learning (EL) approaches have gained popularity in recent years due to their excellent results across various medical domains. This study focuses on their application in disease diagnosis and monitoring. We present a framework designed to provide aspiring researchers with a foundational understanding of popular algorithms and their significance in disease identification. Additionally, we highlight the importance of blockchain technology in the healthcare industry for safeguarding patient data confidentiality and privacy. The decentralized and immutable nature of blockchain can enhance data security, promote interoperability, and empower patients to control their medical information. By demonstrating the potential of advanced ML methods and blockchain technology to transform healthcare systems and improve patient outcomes, our research contributes to the field of disease diagnostics.

Autonomous vehicles (AVs) are poised to become integral components of intelligent transportation systems, particularly within the framework of future smart cities. Traditional performance metrics such as throughput and latency fall short in adequately addressing the temporal relevance and freshness of data in critical applications such as autonomous driving and accident prevention. Consequently, this paper delves into the challenge of reducing the Age of Information (AoI) for disseminating data streams within AV-assisted vehicular networks. Given the dynamic nature of the environment, the problem is formulated as a Markov decision process and tackled using Q-learning and DDQN, both prominent reinforcement learning (RL) algorithms. Additionally, a heuristic approach is introduced to augment the performance of the RL algorithms, expediting environmental learning convergence. The numerical findings underscore the effectiveness of the proposed methodologies in minimizing the aggregate AoI across all data streams.

Non-flat surfaces pose difficulties for robots operating in unstructured environments. Reconstructions of uneven surfaces may only be partially possible due to non-compliant end-effectors and limitations on vision systems such as transparency, reflections, and occlusions. This study achieves blind surface reconstruction by harnessing the robotic manipulator’s kinematic data and a compliant tactile sensing module, which incorporates inertial, magnetic, and pressure sensors. The module’s flexibility enables us to estimate contact positions and surface normals by analyzing its deformation during interactions with unknown objects. While previous works collect only positional information, we include the local normals in a geometrical approach to estimate curvatures between adjacent contact points. These parameters then guide a spline-based patch generation, which allows us to recreate larger surfaces without an increase in complexity while reducing the time-consuming step of probing the surface. Experimental validation demonstrates that this approach outperforms an off-the-shelf vision system in estimation accuracy. Moreover, this compliant haptic method works effectively even when the manipulator’s approach angle is not aligned with the surface normals, which is ideal for unknown non-flat surfaces.

This study presents a comparative analysis of two robotic hand prototypes: one with parallel grippers and the other with flexible grippers featuring articulated finger phalanges. The objective is to determine the most suitable model for manipulating objects in unstructured environments. We identified functional requirements and analyzed data from tactile sensors mounted on the robotic fingertips. We identified objects based on captured grasp and sensor data using artificial intelligence techniques. During pattern recognition exploration, data were collected via serial communication from sensors connected to a microcontroller. Machine learning algorithms with the AutoML technique were used to evaluate the performance of each gripper. The results showed that the hand with parallel grippers outperformed the flexible one in terms of accuracy, recall, precision, F1-score, achieving an average accuracy of 90.95%, compared to 51.10% for the flexible hand. These findings highlight the importance of selecting appropriate robotic grippers for superior object recognition performance.

Vehicle-to-vehicle (V2V) communications retain immense potential in elevating network throughput for next-generation vehicular applications. This study investigates the problem of maximizing the total number of communications while ensuring fairness among V2V communication pairs (MVGCF). In the experiments conducted, each vehicle has a dedicated data stream to be shared among others in the same group. However, not all pairs can directly communicate due to communication range limitations. Hence, the current study focuses on relaying data packets in networks through multi-hop vehicles sharing resource blocks within a time frame while adhering to signal-to-interference-plus-noise ratio (SINR) and half-duplex constraints. To accomplish the research objectives mentioned above, two reinforcement learning (RL) algorithms, namely Q-Learning and Double Deep Q-Networks (DDQN), are proposed. However, to overcome the scalability and computational limitations of RL methods, we devise hybrid heuristic-based reinforcement learning methods, MVGCF_QLearning and MVGCF_DDQN. The numerical results demonstrate the hybrid approaches' effectiveness in terms of the number of successful communications and max-min fairness when compared to a random_agent, and the conventional RL methods for small and large networks.

Maximizing vehicle-to-vehicle (V2V) communications have become crucial for ever-increasing demands for more complicated vehicular applications of smart cities. In this study, the problem of establishing communications between all pairs of vehicles in a group by considering relaying data packets is investigated. The objective is to maximize the total number of communications for groups of vehicles while maintaining fairness among all V2V communication pairs (MCFGV). Reusing resource blocks under the signal-to-interference-plus-noise ratio (SINR) constraint is allowed. We first mathematically formulate the MCFGV problem to find optimal solutions. Then, due to NP-hardness of the problem, we propose a scalable method to solve it for large networks. Finally, through numerical results, the proposed method is compared with the optimum solutions for small networks, and its performance on larger instances is compared to a baseline heuristic.

This paper proposes a set of evolving Takagi-Sugeno-Kang (TSK) fuzzy models of the nonlinear dynamic mechanisms occurring in the myoelectric-based control of prosthetic hand fingers. The rule bases and the parameters of the TSK fuzzy models are evolved by an online identification algorithm. The experimental results prove the performance of the TSK fuzzy models by good output responses and root mean square error values. A performance comparison with two recurrent neural network architectures is included.

This paper proposes a Multilayer Perceptron application in a FastSLAM solution for landmarks update aimed at improve computational performance. The algorithm extracts landmarks from visual sensors, and uses a common features map for two robots. Both activities (i. e. location for the robots and exploration task) are coordinate by central agent. Landmarks update a than are done with a neural network using an Extended Kalman Filter, as usual in several research works in literature. Experiments have shown that generated errors obtained are equivalent in both methods. However, processing time is 10–12 times lower when using our proposed method. This contributes to attend real time requirements during autonomous robot operation.