Embodied agents based on large language models (LLMs) face significant challenges in collaborative tasks, requiring effective communication and reasonable division of labor to ensure efficient and correct task completion. Previous approaches with simple communication patterns carry erroneous or incoherent agent actions, which can lead to additional risks. To address these problems, we propose Cooperative Tree Search (CoTS), a framework designed to significantly improve collaborative planning and task execution efficiency among embodied agents. CoTS guides multi-agents to discuss long-term strategic plans within a modified Monte Carlo tree, searching along LLM-driven reward functions to provide a more thoughtful and promising approach to cooperation. Another key feature of our method is the introduction of a plan evaluation module, which not only prevents agent action confusion caused by frequent plan updates but also ensures plan updates when the current plan becomes unsuitable. Experimental results show that the proposed method performs excellently in planning, communication, and collaboration on embodied environments (CWAH and TDW-MAT), efficiently completing long-term, complex tasks and significantly outperforming existing methods.

Due to the aircraft wing’s topological structure and lightweight design requirements, strain sensors installed on the wing are very limited. Traditional methods, relying on limited sensor data as a single information source, are insufficient for full-stress field monitoring, leading to a high prediction error. To address this issue, a novel wing stress field reconstruction method with limited measurement points is developed via multi-source heterogeneous information fusion. To be specific, two information fusion modules are designed to jointly overcome the challenges of limited measurement data and high non-linearity during full-stress field reconstruction. On one hand, the finite element mechanism-based information fusion module (FEMIFM) is proposed to derive and establish a mechanical model that relates the wing stress to positional parameter, in order to introduce physical information and reduce the non-linearity of the reconstruction mapping. On the other hand, the simulation stress expectation-based information fusion module (SSEIFM) leverages stress expectations derived from simulated stress fields under various operating conditions to incorporate statistical information, thereby enhancing the robustness and reasonableness of reconstruction results. Moreover, a soft-threshold loss function is proposed, which ignores zero-drift errors of strain sensors, improving the reconstruction accuracy of critical stress points. Finally, the developed method can be seamlessly integrated with popular neural networks (i.e., Transformer, convolutional neural networks, multilayer perceptron, etc.). Extensive experiments are conducted to validate the effectiveness of the developed method on an actual aircraft wing stress dataset.

One of the significant tasks in remaining useful life (RUL) prediction is to find a good health indicator (HI) that can effectively represent the degradation process of a system. However, it is difficult for traditional data-driven methods to construct accurate HIs due to their incomprehensive consideration of temporal dependencies within the monitoring data, especially for aeroengines working under nonstationary operating conditions (OCs). Aiming at this problem, this article develops a novel unsupervised deep neural network, the so-called times series memory auto-encoder with sequentially updated reconstructions (SUR-TSMAE) to improve the accuracy of extracted HIs, which directly takes the multidimensional time series as input to simultaneously achieve feature extraction from both feature-dimension and time-dimension. Further, to make full use of the temporal dependencies, a novel long-short time memory with sequentially updated reconstructions (SUR-LSTM), which uses the errors not only from the current memory cell but also from subsequent memory cells to update the output layer's weight of the current memory cell, is developed to act as the reconstructed layer in the SUR-TSMAE. The use of SUR-LSTM can help the SUR-TSMAE rapidly reconstruct the input time series with higher precision. Experimental results on a public dataset demonstrate the outstanding performance of SUR-TSMAE in comparison with some existing methods.

An object-oriented distributed workflow management system was developed. Based on the five essential workflow relations defined by workflow management coalition, a method of common workflow modeling was given. This workflow management system adopted J2EE and JMS technologies to realize the functions of distributed operation and bi-directional visiting between server side and client side. This system has been tested in the workflow of scheme design of large-scale hydraulic generator, and it works steadily and reliably.

A Web-based collaborative manufacturing system has been developed for geographically dispersed team members to conduct design and manufacturing tasks over the Internet. It extends an existing single-location manufacturing system to a multi-location application with the implementation of browser/server computing model and the development tool of active server pages. In this system, a parallel kinematic machine (PKM) is used as the manufacturing tool. The major operations related to the PKM, such as pre-processing, parametric programming, adjustment computing, and cutter motion control, could be conducted online interactively and dynamically. In this paper, the PKM junctions and database of the system are presented, in which the activeX control technology and activeX data object are utilized. Examples are provided to illustrate the latest development of the research.