With the rapid advancement of generative artificial intelligence, embodied intelligence is increasingly being integrated into robotic tasks within smart manufacturing. Industrial operations typically feature long-horizon sequences and demand high precision. However, although embodied models, particularly Vision-Language-Action (VLA) models, have shown significant promise in autonomous control with superior generalization capabilities, their inherent black-box nature lacks the transparency and trustworthiness essential for manufacturing tasks. To address this challenge, we propose the Industrial Embodied Intelligence Trustworthy Interpretable Agent (IEI-TIA), a high-level trustworthy supervisory agent designed for industrial sorting and palletizing tasks. Rather than directly generating low-level control commands, IEI-TIA monitors robotic execution, diagnoses possible failure types at key operational stages, and provides corrective next-step instruction via invoked skills, thereby improving the reliability and interpretability of long-horizon and repetitive robotic tasks. An alignment-enhanced parameter-efficient finetuning method is proposed utilizing paired vision-language data to enable robust trustworthiness identification across various possible robotic failure types in different subtasks in sorting and palletizing. Additionally, we construct a specialized dataset and establish benchmarks for industrial embodied intelligence behavior diagnosis and instructions. Experimental results demonstrate that the fine-tuned trustworthy agent achieves 93.7% diagnosis accuracy for various task failures, representing a substantial improvement of approximately 38% over baseline general-purpose large models. By integrating the agent’s step-by-step operational instructions, the success rate of robotic long-horizon and repetitive tasks is improved by 14.0%.

Grasping tasks are crucial in industrial manufacturing, where precise and efficient object grasping ensures smooth assembly processes and stable operations. Robotic arms, deployed in industrial environments, require timely and accurate computations to perform these tasks. This paper introduces BGRN, an RGB-D fusion-based binary grasp prediction network designed for lightweight grasp pose prediction in such settings. We propose a binary grasp pose prediction framework that significantly reduces memory usage by quantizing both weights and activations to 1 bit. Additionally, an interaction fusion module improves the integration of RGB and depth images, while an information recovery connection helps mitigate feature loss caused by binarization. Experimental results show that BGRN achieves competitive accuracy and notable reductions in memory usage and computational load compared to full-precision models.

An end-to-end compressed sensing (CS) framework, smoothed convolutional CS network with ternary projection (SC-CSNet-TP), has been proposed in this article to deal with problems, including unacceptable cost (i.e., high computational complexity and vast storage requirement), blocking artifact incurred by block CS (BCS), checkerboard artifacts, etc., in order to realize higher reconstruction quality at lower cost. SC-CSNet-TP adopts the framework of CSNet, which consists of a sampling network, an initial reconstruction network, and a deep reconstruction network, and introduces more improvements: 1) sampling is actually a procedure of BCS and can be implemented by convolutions, whose filters form the measurement matrix. In order to balance the precision and hardware pressure, we take ternary measurement matrices into account, which are obtained by pruning binarization process through the self-attention mechanism; 2) the initial reconstruction network focuses on mitigating the blocking artifacts. Blocks restored by inverse convolutions from compressed measurements are smoothed by smoothed projected Landweber (SPL), followed by depth separable convolutions to exploit interblock semantic correlation, with the output upscaled by pixel shuffle to form an initially reconstructed image; and 3) the deep reconstruction network is made of two base blocks interpolated by a dilated convolution, which captures multiscale contextual information to avoid checkerboard artifacts possibly incurred by dilated convolution, and further refines the initial reconstructed image. Experimental results show that the quality of reconstruction by means of SC-CSNet-TP reaches a satisfactory level, e.g., the average peak signal-to-noise ratio on Set11 has approximate 6% improvement compared with that of$\text {DR}^{2}\text {-Net}$when the sampling rate is 0.25.

Recently, foundation models (such as ChatGPT) have emerged with powerful learning, understanding, and generalization abilities, showcasing tremendous potential to revolutionarily promote modern industry. Despite significant advancements in various fields, existing general foundation models face challenges in industry when dealing with the data of specialized modalities, the tasks of varying-scenario with multiple processes, and the requirements of trustworthy output, which makes industrial foundation model (IFM) a necessity. This article proposes a system architecture of termed IFMsys, including model training, model adaptation, and model application. Specifically, in model training, a base model is constructed by pretraining on multimodal industrial data and fine-tuning with fundamental industrial mechanisms. In model adaptation, the base model is developed into a series of task-oriented and domain-specific IFMs through fine-tuning with representative tasks and domain knowledge. In model application, an industrial agent-centric collaboration system and a comprehensive application framework of IFM are proposed to enhance the industrial product lifecycle applications. In addition, a prototype system of the IFM, namely, MetaIndux, is delivered, with application examples presented in typical industrial tasks. Finally, future research directions and open issues of IFM are prospected. We hope this article will inspire the advancements in the theories, technologies, and applications in this emerging research field of IFM.

Complex industrial equipment plays a critical role in specific tasks within industrial edge scenarios. Predicting their health status accurately is essential to ensuring safety and reliability in the production process. However, real-world industrial edge scenarios often have limited resources and stringent real-time requirements, making it difficult to deploy high-precision deep learning models directly at the edge. To address this issue, this article proposes an efficient adaptive-gating binary neural network (ABNN). First, a trend-aware encoder (TAE) is proposed to optimize the binarization process of the input layer. Next, a learnable precision indicator (LPI) is proposed to adjust the inference precision level. Finally, an adaptive-gating convolution is proposed to improve the representational capabilities while maintaining the fitting ability without significantly increasing the computational cost. Additionally, a field-programmable gate array (FPGA) hardware accelerator is designed for the proposed network. ABNN achieves approximately a 7% improvement in accuracy and a 45% gain in efficiency compared to the baseline model.

Edge computing exploits node devices situated in close proximity to terminals to deliver distributed computing services directly to users, with FPGAs serving as the predominant platform. With the amplification in volume and intricacy of deep learning models, effectively deploying these models on FPGA devices introduces substantial challenges. To counter this predicament, this paper introduces an FPGA-based hardware acceleration framework for reconfigurable edge devices tailored for Convolutional Neural Networks (CNNs). Initially, a method for dynamically quantizing network models is devised to markedly curtail model memory consumption. Subsequently, a meticulously engineered hardware acceleration unit is formulated to attain augmented computing parallelism via meticulous temporal redesign. Finally, model deployment on FPGA devices for inference verification is showcased utilizing fully connected and convolutional neural networks as exemplars. On the MNIST dataset, the FPGA inference unit attains remarkable accuracy and computational efficiency in comparison to CPUs and GPUs

Prediction of industrial time-series is crucial for various Industrial Internet of Things applications. Despite the high accuracy of deep learning methods for time-series prediction, the significant memory requirements of deep learning models pose a challenge for the limited computational resources of industrial edge devices. To address this issue, this work proposes BTFormer, which achieves a high compression rate while maintaining competitive performance. First, a binary adaptive attention module is proposed to mitigate the loss of attention information caused by binarization. Second, a trend information soft-link is proposed to propagate trend information between layers and improve the representation ability of the model. Finally, a distribution-guided distillation strategy is proposed to optimize the training process. The experiments demonstrate that BTFormer effectively reduces model memory usage by 31.0 times and improves computational efficiency by 32.8 times while maintaining competitive performance.