The convergence of machine learning (ML) and applied neuroscience continues to accelerate, driven by the synergistic demands of intelligent systems and deepening insights into the human nervous system. Building upon the success of Machine Learning and Applied Neuroscience: Volume I [https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/ fnbot.2023.1191045/full], this second volume brings together cutting-edge research that exemplifies how computational intelligence-particularly deep learning, selfsupervision, and generative modeling-can address complex challenges in neurorobotics, neurorehabilitation, and behavior-aware intelligent systems.The four contributions in this Research Topic span a compelling spectrum: from the diagnosis of gait dysfunction in stroke survivors using cost-sensitive classifiers, to the generation of lifelike 3D human motion through generative adversarial networks (GANs), to next-generation sequential recommendation systems that model multigranularity behavior and feature interactions. Though seemingly diverse, these works share a unifying vision: leveraging advanced ML not only to model neural or behavioral data more accurately, but to extract clinically or functionally meaningful signals that empower real-world applications.One axis of innovation lies in clinical decision support through interpretable and robust ML. In their study, "Machine learning-based gait adaptation dysfunction identification using CMill-based gait data", Yang et al. (2024) (https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/ fnbot.2024.1421401/full) tackle gait adaptation dysfunction (GAD)-a pervasive yet under-assessed impairment in post-stroke patients. Using data from an augmentedreality CMill treadmill, they extract kinematic and adaptability features across four ecologically valid tasks (e.g., obstacle avoidance, slalom walking). Among five classifiers evaluated, the AdaCost algorithm-designed to handle class imbalance and misclassification costs-achieved the best sensitivity (80%) and AUC (0.75). Crucially, feature importance analysis revealed that obstacle avoidance success and gait speed were the top predictors, aligning with clinical intuition and offering actionable biomarkers for rehabilitation planning. This work demonstrates how thoughtful integration of domain-aware data collection and cost-sensitive learning can yield deployable diagnostic aids.Parallel advances emerge in synthetic data generation for human-motion understanding. Wang et al. (2024), entitled "3D human pose data augmentation using Generative Adversarial Networks for robotic-assisted movement quality assessment" (https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/ fnbot.2024.1371385/full) introduce a novel GANs-SVM-DenseNet pipeline to augment 3D human pose datasets-addressing a persistent bottleneck in training data scarcity and limited motion diversity. Their framework uses robotic-assisted capture for highfidelity grounding, GANs to generate realistic and varied motion sequences, DenseNet for hierarchical feature extraction, and SVM for precise motion-quality classification. Evaluated across four benchmarks (Human3.6M, MPI-INF-3DHP, NTU RGB+D, HumanEva), the model outperforms state-of-the-art methods in both accuracy (>96% on Human3.6M) and efficiency (30% faster inference). By closing the loop between data synthesis, feature learning, and quality assessment, this approach paves the way for scalable, robot-in-the-loop systems in sports science, rehabilitation, and virtual reality.Complementing these human-centered applications, two articles push the frontiers of sequential modeling in behavior-aware AI, with implications for neuroscience-inspired user modeling. Zhu et al. (2024a), at "Multi-granularity contrastive learning model for next POI recommendation" (https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/ fnbot.2024.1428785/full), propose MGCL (Multi-Granularity Contrastive Learning) for next Point-of-Interest (POI) recommendation. Recognizing that user mobility is expressed not only at the location level but also through regions and categories, MGCL constructs multi-granular sequences and applies contrastive learning to encourage mutual enhancement across granularities. Experiments on three real-world datasets show consistent gains over 11 baselines-validating that modeling collaborative signals across abstraction levels mitigates data sparsity and enriches preference representation. This principle resonates with hierarchical processing in the brain, where sensory inputs are integrated across spatial and categorical scales.Extending this theme, the same team presents FIDS (Feature Interaction Dual Self-Attention Network) in the article entitled "Feature Interaction Dual Self-attention network for sequential recommendation" (https://www.frontiersin.org/journals/neurorobotics/articles/10.3389/ fnbot.2024.1456192/full), which challenges the assumption that item features are independent. FIDS first uses intra-item self-attention to model higher-order feature interactions (e.g., between brand, category, and seller in e-commerce), then employs dual self-attention streams to capture sequential dependencies both in item sequences and in the derived integrated-feature sequences. This architecture outperforms strong baselines-including FDSA and SASRec-by up to 9% in HR@5 on the Tmall dataset, proving that explicit modeling of feature interdependence enhances behavioral prediction. Such architectures may inform computational models of cognitive binding, where disparate perceptual attributes are fused into coherent percepts.Collectively, these works illustrate a maturing field where ML and neuroscience coevolve: ML architectures grow more neurobiologically plausible (e.g., through hierarchy, attention, and contrastive learning), while neuroscience and clinical applications benefit from increasingly nuanced, robust, and interpretable models.Looking ahead, key challenges remain-particularly around model explainability, realtime deployment in assistive robotics, cross-population generalizability, and ethical handling of behavioral data. Yet the trajectory is clear: the fusion of machine intelligence and applied neuroscience will continue to yield technologies that are not only more intelligent but also more human-centered.

We propose ESSC-RM, a plug-and-play Enhancing framework for Semantic Scene Completion with a Refinement Module, which can be seamlessly integrated into existing semantic scene completion (SSC) models. ESSC-RM operates in two phases: a baseline SSC network first produces a coarse voxel prediction, which is subsequently refined by a 3D U-Net-based Prediction Noise-Aware Module (PNAM) and Voxel-level Local Geometry Module (VLGM) under multiscale supervision. Experiments on SemanticKITTI show that ESSC-RM consistently improves semantic prediction performance. When integrated into CGFormer and MonoScene, the mean IoU increases from 16.87 to 17.27% and from 11.08 to 11.51%, respectively. These results demonstrate that ESSC-RM serves as a general refinement framework applicable to a wide range of SSC models. Project page: https://github.com/LuckyMax0722/ESSC-RM and https://github.com/LuckyMax0722/VLGSSC.

With the advancement of deep learning, road crack segmentation has become increasingly crucial for intelligent transportation safety. Despite notable progress, existing methods still face challenges in capturing fine-grained textures in small crack regions, handling blurred edges and significant width variations, and performing multi-class segmentation. Moreover, the high computational cost of training such models hinders their practical deployment. To tackle these limitations, we propose RSA-TransUNet, a novel model for road crack segmentation. At its core is the Axial-shift MLP Attention (ASMA) mechanism, which integrates axial perception with sparse contextual modeling. Through multi-path axial perturbations and an attention-guided structure, ASMA effectively captures long-range dependencies within row-column patterns, enabling detailed modeling of multi-scale crack features. To improve the model’s adaptability to structural irregularities, we introduce the Adaptive Spline Linear Unit (ASLU), which enhances the model’s capacity to represent nonlinear transformations. ASLU improves responsiveness to microstructural variations, morphological distortions, and local discontinuities, thereby boosting robustness across different domains. We further develop a Structure-aware Multi-stage Evolutionary Optimization (SMEO) strategy, which guides the training process through three phases: structural perception exploration, feature stability enhancement, and global perturbation. This strategy combines breadth sampling, convergence compression, and local escape mechanisms to improve convergence speed, global search efficiency, and generalization performance. Extensive evaluations on the Crack500, CFD, and DeepCrack datasets—including ablation studies and comparative experiments—demonstrate that RSA-TransUNet achieves superior segmentation accuracy and robustness in complex road environments, highlighting its potential for real-world applications.

The underwater environment is more complex than that on land, resulting in severe static and dynamic blurring in underwater images, reducing the recognition accuracy of underwater targets and failing to meet the needs of underwater environment detection. Firstly, for the static blurring problem, we propose an adaptive color compensation algorithm and an improved MSR algorithm. Secondly, for the problem of dynamic blur, we adopt the Restormer network to eliminate the dynamic blur caused by the combined effects of camera shake, camera out-of-focus and relative motion displacement, etc. then, through qualitative analysis, quantitative analysis and underwater target detection on the enhanced dataset, the feasibility of our underwater enhancement method is verified. Finally, we propose a target recognition network suitable for the complex underwater environment. The local and global information is fused through the CCBC module and the ECLOU loss function to improve the positioning accuracy. The FasterNet module is introduced to reduce redundant computations and parameter counting. The experimental results show that the CFEC-YOLOv7 model and the underwater image enhancement method proposed by us exhibit excellent performance, can better adapt to the underwater target recognition task, and have a good application prospect.

Small object detection is a critical task in applications like autonomous driving and ship black smoke detection. While Deformable DETR has advanced small object detection, it faces limitations due to its reliance on CNNs for feature extraction, which restricts global context understanding and results in suboptimal feature representation. Additionally, it struggles with detecting small objects that occupy only a few pixels due to significant size disparities. To overcome these challenges, we propose the Context-Aware Enhanced Feature Refinement Deformable DETR, an improved Deformable DETR network. Our approach introduces Mask Attention in the backbone to improve feature extraction while effectively suppressing irrelevant background information. Furthermore, we propose a Context-Aware Enhanced Feature Refinement Encoder to address the issue of small objects with limited pixel representation. Experimental results demonstrate that our method outperforms the baseline, achieving a 2.1% improvement in mAP.

Reliable visual perception is essential for autonomous driving test scenario generation, yet adverse weather and lighting variations pose significant challenges to simulation robustness and generalization. Traditional unpaired image-to-image translation methods primarily rely on RGB-based transformations, often resulting in geometric distortions and loss of structural consistency, which can negatively impact the realism and accuracy of generated test scenarios. To address these limitations, we propose a Depth-Aware Dual-Branch Generative Adversarial Network (DAB-GAN) that explicitly incorporates depth information to preserve spatial structures during scenario generation. The dual-branch generator processes both RGB and depth inputs, ensuring geometric fidelity, while a self-attention mechanism enhances spatial dependencies and local detail refinement. This enables the creation of realistic and structure-preserving test environments that are crucial for evaluating autonomous driving perception systems, especially under adverse weather conditions. Experimental results demonstrate that DAB-GAN outperforms existing unpaired image-to-image translation methods, achieving superior visual fidelity and maintaining depth-aware structural integrity. This approach provides a robust framework for generating diverse and challenging test scenarios, enhancing the development and validation of autonomous driving systems under various real-world conditions.

Robotic systems have significantly evolved, integrating Artificial Intelligence (AI) techniques to enhance autonomy, adaptability, and robustness. Traditional control methods, such as Proportional-Integral-Derivative (PID) controllers, remain fundamental in many applications, yet AI-driven approaches are proving invaluable in tackling complex, non-linear, and uncertain environments.This special issue explores the latest advancements in AI-driven control techniques for robotic applications. The contributions cover a range of topics, including adaptive and learning-based control strategies, intelligent navigation, state estimation, and the integration of hybrid AI techniques for enhanced robotic performance.The articles included in this special issue address various challenges in robotic control and introduce innovative AI-based methodologies. Below, we highlight the key contributions:• Intelligent Control Techniques: Several articles explore the integration of AI-driven controllers, such as neural networks and fuzzy logic, to improve robotic decision-making and adaptability in dynamic environments.• State Estimation and Sensor Fusion: Advanced techniques like Kalman filters and particle filters are leveraged to enhance real-time localization and trajectory tracking, ensuring precise motion planning.• Hybrid Control Strategies: The fusion of classical and AI-based methods demonstrates how hybrid controllers can achieve robust performance, particularly in unstructured and uncertain scenarios.• Obstacle Avoidance and Path Optimization: Contributions discuss AI-enhanced navigation algorithms that enable robots to autonomously detect obstacles and compute optimal paths in real time.The first paper, "On Designing a Configurable UAV Autopilot for Unmanned Quadrotors," presents a control strategy for unmanned aerial vehicles (UAVs) based on an inner-outer loop design for attitude and altitude control. It leverages nonlinear feedback linearization, linear-quadratic regulator (LQR), sliding mode control (SMC), proportional-derivative (PD), and PID controllers to enhance stability and disturbance rejection. The study also introduces a PD-PID hybrid controller designed for tracking and surveillance of smoke and fire. Simulation results confirm the effectiveness of the proposed approach in improving UAV performance under various conditions. The second paper, "Fusion Inception and Transformer Network for Continuous Estimation of Finger Kinematics from Surface Electromyography (sEMG)," introduces FIT (Fusion Inception Transformer), a deep learning model for decoding sEMG signals to predict hand movements. By integrating Inception and Transformer networks, the model efficiently extracts both local and global features. Results on the Ninapro dataset show that FIT outperforms temporal convolutional networks (TCN), long short-term memory (LSTM) networks, and Bidirectional Encoder Representations from Transformers (BERT) in terms of estimation accuracy and computational efficiency, contributing significantly to advancements in human-computer interaction (HCI) technology.The third paper, "Cardioid Oscillator-Based Pattern Generator for Imitating Lower Limb Exoskeleton Behavior," proposes a pattern generator based on cardioid oscillators to mimic human gait characteristics such as periodicity, selfexcitation, and time-ratio asymmetry. The approach is validated through simulations and experiments, demonstrating that the generated trajectories closely resemble natural human gait. This method holds promise for exoskeleton control in rehabilitation and assistive applications.The fourth paper, "An Autonomous Mobile Robot Path Planning Strategy Using an Enhanced Slime Mold Algorithm (ESMA)," introduces ESMA, an improved optimization method for robotic path planning. By incorporating adaptive techniques for enhanced global search and an artificial potential field for dynamic obstacle avoidance, ESMA significantly reduces both path length and computation time. Comparative analysis against the standard Slime Mold Algorithm (SMA) and other optimization algorithms confirms its superiority, making it a valuable solution for real-world robotic navigation.The fifth paper, "Multi-User Motion Recognition Using sEMG via Discriminative Canonical Correlation Analysis and Adaptive Dimensionality Reduction," proposes a multi-user sEMG motion recognition framework designed to address variability in muscle signals across different users. By utilizing discriminative canonical correlation analysis (DCCA) and adaptive dimensionality reduction (ADR), the system projects feature sets into a uniform space, enhancing motion recognition accuracy. Experimental results demonstrate over 90% accuracy, highlighting its potential for rehabilitation and assistive technology applications.The convergence of AI and robotics presents exciting research opportunities. Future advancements may focus on reinforcement learning for real-time adaptation, neuromorphic computing for energy-efficient robotic control, and explainable AI (XAI) for transparent decision-making in safety-critical applications.This special issue highlights the growing role of AI in robotic applications and control. The featured articles provide valuable insights into emerging methodologies that push the boundaries of autonomous robotic systems. We thank all the contributors for their efforts in advancing this field.We express our gratitude to the authors, reviewers, and editorial team for their contributions to this special issue. Their efforts have ensured a high-quality collection of research in AI-driven robotics.

The integration of neural network models in autonomous robotics represents a monumental leap in artificial intelligence and robotics. These models, mirroring the human brain's complexity and efficiency, have catalyzed innovations in machine learning, fostering more adaptive, intelligent, and efficient robotic systems. Recent research in areas like deep learning, reinforcement learning, and neural network optimization has significantly advanced, yet challenges remain, especially in robotics' real-world application, energy efficiency, and operation in complex, unstructured environments.

The development of air traffic control (ATC) automation has been constrained by the scarcity and low quality of communication data, particularly in low-altitude complex airspace, where non-standardized instructions frequently hinder training efficiency and operational safety. This paper proposes the BART-Reinforcement Learning (BRL) model, a deep reinforcement learning model based on the BART pre-trained language model, optimized through transfer learning and reinforcement learning techniques. The model was evaluated on multiple ATC datasets, including training flight data, civil aviation operational data, and standardized datasets generated from Radiotelephony Communications for Air Traffic Services. Evaluation metrics included ROUGE and semantic intent-based indicators, with comparative analysis against several baseline models. Experimental results demonstrate that BRL achieves a 10.5% improvement in overall accuracy on the training dataset with the highest degree of non-standardization, significantly outperforming the baseline models. Furthermore, comprehensive evaluations validate the model's effectiveness in standardizing various types of instructions. The findings suggest that reinforcement learning-based approaches have the potential to significantly enhance ATC automation, reducing communication inconsistencies, and improving training efficiency and operational safety. Future research may further optimize standardization by incorporating additional contextual factors into the model.

Aiming at the problems of slow network convergence, poor reward convergence stability, and low path planning efficiency of traditional deep reinforcement learning algorithms, this paper proposes a BiLSTM-D3QN (Bidirectional Long and Short-Term Memory Dueling Double Deep Q-Network) path planning algorithm based on the DDQN (Double Deep Q-Network) decision model. Firstly, a Bidirectional Long Short-Term Memory network (BiLSTM) is introduced to make the network have memory, increase the stability of decision making and make the reward converge more stably; secondly, Dueling Network is introduced to further solve the problem of overestimating the Q-value of the neural network, which makes the network able to be updated quickly; Adaptive reprioritization based on the frequency penalty function is proposed. Experience Playback, which extracts important and fresh data from the experience pool to accelerate the convergence of the neural network; finally, an adaptive action selection mechanism is introduced to further optimize the action exploration. Simulation experiments show that the BiLSTM-D3QN path planning algorithm outperforms the traditional Deep Reinforcement Learning algorithm in terms of network convergence speed, planning efficiency, stability of reward convergence, and success rate in simple environments; in complex environments, the path length of BiLSTM-D3QN is 20 m shorter than that of the improved ERDDQN (Experience Replay Double Deep Q-Network) algorithm, the number of turning points is 7 fewer, the planning time is 0.54 s shorter, and the success rate is 10.4% higher. The superiority of the BiLSTM-D3QN algorithm in terms of network convergence speed and path planning performance is demonstrated.