Frontiers in Neurorobotics最新文献

Balancing exploration and exploitation remains a fundamental challenge in reliable mobile robot control, as conventional policies often converge on suboptimal behaviors. Inspired by the brain's division of labor for adaptive control, we propose SpikeAEC, a fully spiking, neuromodulated Actor-Explorer-Critic architecture designed to address this dilemma online within a closed-loop system. SpikeAEC comprises three specialized subnetworks operating in parallel: the Actor, inspired by the basal ganglia, proposes exploitative actions; the Explorer, modeled after the ACC-GPe-STN pathway, generates adaptive exploratory actions gated by a vigilance signal modulated by the accumulated global temporal-difference (TD) error; and the Critic, based on the ventral striatum, computes the TD error. The final action is selected by a separate, TAN-based Arbitrator, which probabilistically chooses between the Actor's and Explorer's action proposals according to recent performance and the TD error. These subnetworks are coupled through a unified three-factor learning framework that uses the TD signal and phasic neuromodulators (acetylcholine and dopamine) from the Arbitrator to drive pathway-specific synaptic plasticity. This online plasticity enhances the quality of action proposals and accelerates policy refinement. In simulation, SpikeAEC outperforms leading brain-inspired methods by converging 24% faster, reducing trajectory length by 18%, and increasing cumulative reward by over 5% against the top-performing baseline, all while maintaining consistency with established neurophysiological principles.

As automotive manufacturing advances toward the industrial 5.0 era, traditional rigid automation production models are transitioning toward the embodied intelligence paradigm. Confronted with mass customization, diverse products, and small-batch production, the environment of automotive manufacturing exhibits high dynamism and unstructured characteristics. Different from traditional industrial intelligence based on static, hard-coded logic, robots enhance their cognitive abilities through closed-loop interaction with dynamic environments, inspired by bionic neural mechanisms, this shift enables robots to perform flexible and reliable operations in complex production scenarios. This paper analyzes the core role and key technologies of neural intelligence algorithms in reshaping perception, decision, and execution of industrial robot, while providing a systematic review of industrial robot evolution within the automotive industry, and provides a reliable path for future development.

Problem: Deep learning technology promotes the development of single-image dehazing. However, many existing methods fail to fully consider the haze density and its spatial distribution, which limits the improvement of dehazing performance.

Proposed solution: To address this issue, we propose an attention-based multi-scale feature aggregation network (AMSA-Net) for single-image dehazing.

Method: AMSA-Net is an encoding and decoding structure. Its encoder and decoder are composed of multi-scale hybrid attention feature aggregation module (MSHA-FAM). The module can perceive the haze density and spatial information in the haze image, which helps to improve the dehazing effect. MSHA-FAM is composed of two key components: the scale-aware coordinate residual module (SCRM) and multi-scale feature refinement residual module (MSFRRM). SCRM uses improved coordinate attention to effectively capture haze density and spatial characteristics, thus significantly improving dehazing effect. MSFRRM extracts semantic features through up-sampling and down-sampling, and uses improved pixel attention mechanism to enhance key features. In the overall MSHA-FAM pipeline, SCRM first learns the density and spatial distribution characteristics of haze, then refines it through MSFRRM, so as to remove haze more effectively.

Key results: The experimental results demonstrate that our proposed AMSA-Net is superior to the comparison methods in terms of dehazing quality. Ablation studies further verify the effectiveness of the proposed modules.

Impact: In this work, we present AMSA-Net, which has achieved good dehazing performance and can provide high-quality input for subsequent computer vision tasks.

Emotion estimation is a field that has been studied for a long time, and several approaches using machine learning models exist. This article presents BlendFER-Lite, an LSTM model that uses Blendshapes from the MediaPipe library to analyze facial expressions detected from a live-streamed camera feed. This model is trained on the FER2013 dataset and achieves 71% accuracy and an F1-score of 62%, meeting the accuracy benchmark for the FER2013 dataset while significantly reducing computational costs compared to current methods. For the sake of reproducibility, the code repository, datasets, and models proposed in this paper, in addition to the preprint, can be found on Hugging Face at: https://huggingface.co/papers/2501.13432.

Jel classification: D8, H51.

Msc classification: 35A01, 65L10, 65L12, 65L20, 65L70.

Introduction: The integration of virtual simulation with intelligent modeling is crucial for advancing the scientization and personalization of volleyball physical training. This study aims to overcome the convergence instability and feature misalignment in modeling multimodal kinematic and physiological sequences.

Methods: A dynamical framework based on a Dual-Stream Long Short-Term Memory network integrated with a temporal attention mechanism is proposed. The framework decouples heterogeneous feature learning and optimizes temporal weight distribution.

Results: Experimental validation on complex motion state estimation demonstrates that the proposed model reduces load modeling error to 3.8% and achieves a motion classification accuracy of 93.1%. The velocity trajectory fitting coefficient of determination is 0.91 with a peak deviation of 0.05 m/s.

Discussion: These results confirm the effectiveness of the attention-based DS-LSTM in optimizing multimodal sequence modeling for training state estimation and feedback.

Introduction: Telepresence robots (TPRs) must co-navigate with humans in constrained hospital environments, where safety depends on anticipating rather than merely reacting to human motion. Existing approaches rarely integrate short-horizon human-motion forecasting with safety-constrained control, which reduces robustness in dense corridors and ward bays. This study addresses this gap by evaluating an anticipatory, safety-aware co-navigation framework for TPRs.

Methods: We developed a modular framework that couples a lightweight transformer-based forecaster that predicts multi-agent trajectories under occlusion with a safe reinforcement learning (RL) controller. The forecaster produces short-term distributions over pedestrian states that are embedded into the RL policy state and cost as risk-aware occupancy features. Safety is enforced via constrained policy optimization augmented by a run-time control barrier function (CBF) shield that filters unsafe actions. We benchmarked the approach against a social-force or dynamic window approach (DWA), an attention-based crowd-RL policy, and model predictive control (MPC) with CBF. Experiments were conducted across two hospital-like benchmarks (a crowded corridor and a four-bed ward), totaling 2,400 episodes. Outcomes included task success, collision count, minimum human-robot clearance, near-miss events ( ≤ 0.3 m), time-to-goal, CBF violations, and ablations removing forecasting and the CBF shield.

Results: Relative to the best-performing baseline, the proposed method improved task success by 21.6% and reduced collisions by 47.3%. Median minimum human-robot clearance increased by 0.19 m, and near-miss events decreased by 38.5%. Time-to-goal was maintained within +2.7% of MPC+CBF while incurring zero CBF violations under the shield. Ablation studies showed that removing forecasting degraded success by 14.2%, whereas removing the CBF shield increased constraint breaches from 0% to 6.1% of steps.

Discussion: Anticipatory perception combined with Safe-RL yields substantially safer and more reliable telepresence co-navigation in human-dense clinical layouts without sacrificing efficiency. The framework is modular, enabling alternative forecasters and safety shields. Limitations include sensitivity to forecast drift during abrupt changes in crowd flow. Future work will explore on-device adaptation, shared-autonomy overlays to incorporate operator intent, and prospective evaluations in live hospital workflows.

Digital-image technology has broadened the creative space of dance, yet accurately capturing the semantic correspondence between low-level motion data and high-level dance key-points remains challenging, especially when labeled data are scarce. We aim to establish a lightweight, semi-supervised pipeline that can extract discriminative motion features from depth sequences and map them to 3-D key-points of dancers in real time. To achieve pixel-level alignment between dance movement targets and high-dimensional sensory data, we propose a novel LSTM-CNN (Long Short Term Memory-Convolutional Neural Network) framework. Temporal-context features are first extracted by LSTM, after which multi-dimensional spatial features are captured by three convolutional layers and one max-pooling layer; the fused representation is finally regressed to 3-D body key-points. To relieve class imbalance caused by complex postures, an online hard-example mining (OHEM) strategy together with a Dice-cross-entropy weighted loss (3:1) is embedded into semi-supervised learning, enabling the network to converge with only 20% labeled samples. Experiments on the public MSR-Action3D dataset (567 sequences, 20 actions) yielded an average recognition rate of 96.9%, surpassing the best comparison method (MSST) by 1.1%. On our self-established dataset (99 sequences, 11 actions) the accuracy reached 97.99% while training time was reduced by 35% compared with the previous best Multi_perspective_MHPCs approach. Both datasets show low RMSE (≤ 0.032) between predicted and ground-truth key-points, confirming spatial precision. The results demonstrate that the proposed model can reliably track subtle dance gestures under limited annotation, offering an efficient, low-cost solution for digital choreography, motion-style transfer and interactive stage performance.

Prosthetic knee joints are essential assistive technologies designed to replicate natural gait and improve mobility for individuals with lower-limb loss. This study presents a comprehensive nonlinear dynamic model of a two-degree-of-freedom prosthetic knee joint and introduces three robust nonlinear control strategies: Integral Sliding Mode Control, Conditional Super-Twisting Sliding Mode Control, and Conditional Adaptive Positive Semidefinite Barrier Function-based Sliding Mode Control. These controllers are designed to address the challenges associated with nonlinear joint dynamics, external disturbances, and modeling uncertainties during locomotion. To optimize control performance, the gain parameters of each controller were fine-tuned using Red Fox Optimization, a metaheuristic algorithm inspired by the intelligent hunting behavior of red foxes. Stability analysis is conducted using Lyapunov theory, and control effectiveness is evaluated through simulations in MATLAB/Simulink and validated via hardware-in-the-loop testing using a C2000 Delfino F28379D microcontroller. Among the three controllers, the CoBA-based approach demonstrated the highest tracking accuracy, fastest convergence, and smoothest torque profile. The close agreement between simulation and experimental results confirms the practical applicability of the proposed control framework, offering a promising solution for intelligent and adaptive prosthetic knee systems.

Motor imagery brain-computer interface (MI-BCI) has garnered considerable attention due to its potential for neural plasticity. However, the limited number of MI-EEG samples per subject and the susceptibility of features to noise and artifacts posed significant challenges for achieving high decoding performance. To address this problem, a Data-Augmented Multi-Scale Temporal Attention Convolutional Network (AMANet) was proposed. The network mainly consisted of four modules. First, the data augmentation module comprises three steps: sliding-window segmentation to increase sample size, Common Spatial Pattern (CSP) extraction of discriminative spatial features, and linear scaling to enhance network robustness. Then, multi-scale temporal convolution was incorporated to dynamically extract temporal and spatial features. Subsequently, the ECA attention mechanism was integrated to realize the adaptive adjustment of the weights of different channels. Finally, depthwise separable convolution was utilized to fully integrate and classify the deep extraction of temporal and spatial features. In 10-fold cross-validation, the results show that AMANet achieves classification accuracies of 84.06 and 85.09% on the BCI Competition IV Datasets 2a and 2b, respectively, significantly outperforming baseline models such as Incep-EEGNet. On the High-Gamma dataset, AMANet attains a classification accuracy of 95.48%. These results demonstrate the excellent performance of AMANet in motor imagery decoding tasks.