Explore the vast range of titles available.

By mimicking the neurons and synapses of the human brain and employing spiking neural networks on neuromorphic chips, neuromorphic computing offers a promising energy-efficient machine intelligence. How to borrow high-level brain dynamic mechanisms to help neuromorphic computing achieve energy advantages is a fundamental issue. This work presents an application-oriented algorithm-software-hardware co-designed neuromorphic system for this issue. First, we design and fabricate an asynchronous chip called “Speck”, a sensing-computing neuromorphic system on chip. With the low processor resting power of 0.42mW, Speck can satisfy the hardware requirements of dynamic computing: no-input consumes no energy. Second, we uncover the “dynamic imbalance” in spiking neural networks and develop an attention-based framework for achieving the algorithmic requirements of dynamic computing: varied inputs consume energy with large variance. Together, we demonstrate a neuromorphic system with real-time power as low as 0.70mW. This work exhibits the promising potentials of neuromorphic computing with its asynchronous event-driven, sparse, and dynamic nature. Mimicking high-level abstraction of the brain to achieve energy advantages is a fundamental issue in neuromorphic computing. Here, the authors fabricate an asynchronous chip and demonstrate a high-accuracy neuromorphic system with power consumption of 0.7mW.

4D millimeter-wave radar provides high-precision ranging capability and exhibits strong robustness under adverse weather and low-visibility conditions, but its point clouds are relatively sparse and suffer from severe elevation-angle measurement noise. Monocular cameras, by contrast, provide rich semantic information and high recall, yet are fundamentally limited by scale ambiguity. To exploit the complementary characteristics of these two sensors, this paper proposes a radar-camera fusion 3D multi-object tracking framework that does not rely on complex 3D annotated data. First, on the radar signal-processing side, a Gaussian distribution-based adaptive angle compression method and IMU-based velocity compensation are introduced to effectively suppress measurement noise, and an improved DBSCAN clustering scheme with recursive cluster splitting and historical static-box guidance is employed to generate high-quality radar detections. Second, a disparity-domain metric depth recovery method is proposed. This method uses filtered radar points as sparse metric anchors, performs robust fitting with RANSAC, and applies Kalman filtering for temporal smoothing, thereby converting the relative depth output of the visual foundation model Depth Anything V2 into metric depth. Finally, a hierarchical fusion strategy is designed at both the detection and tracking levels to achieve stable cross-modal state association. Experimental results on a self-collected dataset show that the proposed method achieves an overall MOTA of 77.93%, outperforming single-modality baselines and other comparison methods by 11 to 31 percentage points. This study provides an effective solution for low-cost and robust environment perception in complex dynamic scenarios.

In the context of large-scale renewable energy integration driven by China’s dual-carbon goals, and under distribution network scenarios with continuously increasing shares of wind and photovoltaic generation, this paper proposes a source–grid–load–storage coordinated planning method embedded with a multi-agent game mechanism. First, the interest transmission pathways among distributed generation operators (DGOs), distribution network operators (DNOs), energy storage operators (ESOs), and electricity users are mapped, based on which a profit model is established for each stakeholder. Building on this, a coordinated planning framework for active distribution networks (DN) is developed under the assumption of bounded rationality. Through an evolutionary-game process among DGOs, DNOs, and ESOs, and in combination with user-side demand response, the model jointly determines the optimal network reinforcement scheme as well as the optimal allocation of distributed generation (DG) and energy storage system (ESS) resources. Case studies are then conducted to verify the feasibility and effectiveness of the proposed method. The results demonstrate that the approach enables coordinated planning of DN, DG, and ESS, effectively guides users to participate in demand response, and improves both planning economy and renewable energy accommodation. Moreover, by explicitly capturing the trade-offs among multiple stakeholders through evolutionary-game interactions, the planning outcomes align better with real-world operational characteristics.

The increased share of new energy sources in Northeast China’s power mix has strained grid stability. Energy storage technologies are essential for maintaining grid stability by addressing peak shaving and frequency regulation challenges. However, a clear quantitative assessment of the region’s energy storage needs is lacking, leading to weak grid stability and limited growth potential. This paper analyzes power supply data from Northeast China and models the stochastic characteristics of new energy generation. A joint optimization model for energy storage and thermal power is developed to optimize power allocation for peak shaving and frequency regulation at minimal cost. The empirical distribution method quantifies the relationship between storage power, capacity, and confidence levels, providing insights into the region’s future energy storage demands. The study finds that under 10 typical scenarios, the demand for peaking power at a 15 min scale is ≤500 MW, and the demand for frequency regulation at a 1 min scale is ≤1000 MW. At the 90% confidence level, the required capacity for new energy storage for peak shaving and frequency regulation is 424.13 MWh and 197.65 MWh, respectively. The required power for peak shaving and frequency regulation is 247.88 MW and 527.33 MW, respectively. The durations of peak shaving and frequency regulation are 1.71 h and 0.38 h. It also forecasts the energy storage capacity in the northeast region from 2025 to 2030 under the 5% annual incremental new energy penetration scenario. These findings provide theoretical support for energy storage policies in Northeast China during the 14th Five-Year Plan and practical guidance for accelerating energy storage industrialization.

It is essential and forward-thinking to investigate the personalized use of four-wheel driving and steering wire-controlled unmanned chassis. This paper introduces a personalized path-tracking approach designed to adapt the vehicle’s control system to human-like characteristics, enhancing the fit and maximizing the potential of the chassis’ multi-directional driving and steering capabilities. By modifying the classic vehicle motion controller design, this approach aligns with individual driving habits, significantly improving upon traditional path-tracking control methods that rely solely on reference vector fields. First, the classic reference vector field’s logic was expanded upon, and it is shown that a personalized upgrade is feasible. Then, driving behavior data from multiple drivers were collected using a driving simulator. The fuzzy c-means clustering method was used to categorize drivers based on typical states that match vehicle path-tracking performance. Additionally, the random forest algorithm was used as the method for recognizing driving style. Subsequently, a personalized path-tracking control strategy based on the reference vector field was developed and a distributed execution architecture for four-wheel driving and steering wire-controlled unmanned chassis was established. Finally, the proposed personalized path-tracking approach was validated using a driving simulator. The results of the experimental tests demonstrated that the personalized path-tracking control approach not only fits well with various driving styles but also delivers high accuracy in driving style identification, making it highly suitable for application in four-wheel driving and steering wire-controlled chassis.

Against the backdrop of automobile electrification, an increasing number of battery-swapping stations for electric vehicles have been launched to address the issue of slow battery charging under cold temperature conditions. However, due to the separation of the discharging and charging processes for lithium-ion batteries (LIBs) at swapping stations, and the circulation of batteries across different vehicles and stations, the operating data become fragmented, making it difficult to accurately identify the battery state-of-health (SOH). This study proposes a BiLSTM-Transformer framework that extracts the Constant Voltage Time (CVT) feature using only charging data, enabling the precise estimation of battery capacity degradation. Validation experiments conducted on battery samples under different operating temperatures showed that the model achieved a normalized RMSE of less than 1.6%. In ideal conditions, the normalized RMSE of the estimation reached as low as 0.11%. This model enables SOH estimation without relying on discharge data, contributing to the efficient and safe operation of battery swapping stations.

In the realm of image and vision computing, shape recognition remains challenging due to intra-class variations that can degrade performance. To address this, we propose an enhanced shape recognition method leveraging dual constraint reduction and adaptive orthogonal wavelet transform (ADCR). Firstly, we refine shape contours by reducing salient points to strong feature points (bone points), mitigating the impact of shape deformation. Secondly, we employ an adaptive orthogonal wavelet transform to ensure scale and mirror invariance, significantly accelerating recognition speed. Thirdly, we streamline the feature structure after wavelet transformation, retaining only minimum, mean, and maximum eigenvectors for efficient shape representation. Experimental results on multiple datasets demonstrate the effectiveness of our method, achieving a MAP score of 79.2% on the ICL dataset with an average recognition time of 26.9 ms, outperforming similar methods by over 200 times in efficiency. Our ADCR method significantly enhances shape recognition capabilities.

The field of trajectory tracking control for ground vehicles has problems such as unknown control object manipulation characteristics, missing dynamic digital models, and susceptibility to distortion of upper-layer tracking control algorithms. This paper first proposes a method for identifying the chassis forward/inverse dynamics model based on vehicle-level state variables and control variable methods to solve these issues. Secondly, the control object manipulation characteristics and dynamic fitting digital model are obtained based on a specific electric drive-tracked chassis. Thirdly, a linear quadratic regulator is used as the core path tracking algorithm, and a dual closed-loop trajectory tracking control algorithm based on the inverse dynamics fitting model is established for both lateral and longitudinal position feedforward + velocity feedback correction. Finally, systematic verification is conducted using the Simulink simulation environment and real vehicles. The verification results show that the proposed model identification method supports vehicle-level state fitting accuracy maintained within 3%, and the trajectory tracking control algorithm maintains tracking accuracy within 0.1m under extreme U-turn conditions. The dual closed-loop control structure enhances control accuracy and robustness.

The increase in the penetration rate of renewable energy such as wind power and photovoltaics has brought challenges to the stable dispatch and safe operation of the power system. Therefore, based on the short-time fast Fourier transform method, this paper constructs a long-short-term memory neural network model, and proposes a new volatility clustering algorithm for intermittent renewable energy. On this basis, based on the open source data set RTS-GMLC published by the National Renewable Energy Laboratory (NREL), experimental simulations were carried out to verify the effectiveness of the proposed volatility clustering algorithm for intermittent renewable energy. sex. This provides a quantitative decision-making basis for the dispatch and operation of power systems dominated by intermittent renewable energy.

Deep neural networks (DNNs) have enabled impressive breakthroughs in various artificial intelligence (AI) applications recently due to its capability of learning high-level features from big data. However, the current demand of DNNs for computational resources especially the storage consumption is growing due to that the increasing sizes of models are being required for more and more complicated applications. To address this problem, several tensor decomposition methods including tensor-train (TT) and tensor-ring (TR) have been applied to compress DNNs and shown considerable compression effectiveness. In this work, we introduce the hierarchical Tucker (HT), a classical but rarely-used tensor decomposition method, to investigate its capability in neural network compression. We convert the weight matrices and convolutional kernels to both HT and TT formats for comparative study, since the latter is the most widely used decomposition method and the variant of HT. We further theoretically and experimentally discover that the HT format has better performance on compressing weight matrices, while the TT format is more suited for compressing convolutional kernels. Based on this phenomenon we propose a strategy of hybrid tensor decomposition by combining TT and HT together to compress convolutional and fully connected parts separately and attain better accuracy than only using the TT or HT format on convolutional neural networks (CNNs). Our work illuminates the prospects of hybrid tensor decomposition for neural network compression.