We present Chain-of-Action (CoA), a novel visuomotor policy paradigm built upon Trajectory Autoregressive Modeling. Unlike conventional approaches that predict next step action(s) forward, CoA generates an entire trajectory by explicit backward reasoning with task-specific goals through an action-level Chain-of-Thought (CoT) process. This process is unified within a single autoregressive structure: (1) the first token corresponds to a stable keyframe action that encodes the task-specific goals; and (2) subsequent action tokens are generated autoregressively, conditioned on the initial keyframe and previously predicted actions. This backward action reasoning enforces a global-to-local structure, allowing each local action to be tightly constrained by the final goal. To further realize the action reasoning structure, CoA incorporates four complementary designs: continuous action token representation; dynamic stopping for variable-length trajectory generation; reverse temporal ensemble; and multi-token prediction to balance action chunk modeling with global structure. As a result, CoA gives strong spatial generalization capabilities while preserving the flexibility and simplicity of a visuomotor policy. Empirically, we observe that CoA outperforms representative imitation learning algorithms such as ACT and Diffusion Policy across 60 RLBench tasks and 8 real-world tasks.

In recent years, numerous studies have sought to enhance the capabilities of pretrained language models (PLMs) for Knowledge Graph Completion (KGC) tasks by integrating structural information from knowledge graphs.However, existing approaches have not effectively combined the structural attributes of knowledge graphs with the textual descriptions of entities to generate robust entity encodings.To address this issue, this paper proposes Mo-CoKGC (Momentum Contrast Entity Encoding for Knowledge Graph Completion), which incorporates three primary encoders: the entityrelation encoder, the entity encoder, and the momentum entity encoder.Momentum contrastive learning not only provides more negative samples but also allows for the gradual updating of entity encodings.Consequently, we reintroduce the generated entity encodings into the encoder to incorporate the graph's structural information.Additionally, MoCoKGC enhances the inferential capabilities of the entity-relation encoder through deep prompts of relations.On the standard evaluation metric, Mean Reciprocal Rank (MRR), the MoCoKGC model demonstrates superior performance, achieving a 7.1% improvement on the WN18RR dataset and an 11% improvement on the Wikidata5M dataset, while also surpassing the current best model on the FB15k-237 dataset.Through a series of experiments, this paper thoroughly examines the role and contribution of each component and parameter of the model.

Denoising plays a vital role in freeform structured surface metrology. Traditional techniques, such as Gaussian and partial differential equation-based diffusion filters, often involve a time-consuming calibration process, particularly for complex surfaces. The main challenge lies in automating the denoising operation while accurately preserving features for varied surface textures. To address this challenge, an automatic approach PI-DnCNN based on small sample learning is presented in this paper. Denoising convolutional neural network (DnCNN) is employed as the basic architecture of this approach, due to its effectiveness in tackling mix-level Gaussian noise and adapting to small training datasets. Acknowledging the constraints of limited datasets, a novel physics-informed denoising loss function marrying filtering techniques is proposed to improve model performance. Additionally, a hybrid data augmentation strategy is developed to enhance the recognition of complex components. The paper also reports a set of experiments to demonstrate the presented approach in terms of performance over conventional techniques, enhancements with limited sample sizes, and applicability in general image denoising. The experiment results suggest that the presented approach consistently achieves higher average scores compared to traditional filters and emerges superior compared to the conventional DnCNN loss across different dataset sizes. In addition, the proposed loss also shows effectiveness in general image denoising, which suggests the robustness and universality of the approach.

In design for additive manufacturing, an essential task is to determine the optimal build orientation of a part according to one or multiple factors. Heuristic search is used by the most part orientation methods to select the optimal orientation from a large solution space. Search algorithms occasionally converge towards the local optimum and waste considerable time on trial and error. This issue could be addressed if there was an intelligent agent that knew the optimal search path for a given 3D model. A straightforward method to construct such an agent is reinforcement learning (RL). By adopting this idea, the time-consuming online searches in existing part orientation methods will be moved to the offline learning stage, potentially improving part orientation performance. This is a challenging problem because the goal is to build an agent capable of rotating arbitrary 3D models, whereas RL agents frequently struggle to generalize in new scenarios. Therefore, this paper suggests a generalizable reinforcement learning (GRL) framework to train the agent, and a GRL benchmark to support the training, testing, and comparison of part orientation approaches. Experimental results de- monstrate that the proposed method on average outperforms others in terms of effectiveness and efficiency. It is proved to have the potential to solve the local minima problems raised in the existing approaches, to swiftly discover the global (sub-)optimal solution (i.e., on average 2.62× to 229.00× faster than the random search algorithm), and to generalize beyond the environment in which it was trained.

In Industrie 4.0, machines are expected to become autonomous, self-aware and self-correcting. One important step in the area of manufacturing is feature recognition that aims to detect all the machining features from a 3-D model. In this research area, recognizing and locating a wide variety of highly intersecting features are extremely challenging as the topology information of features is substantially damaged because of the feature intersection. Motivated by the single shot multibox detector (SSD), this article presents a novel deep learning approach named SsdNet to tackle the machining feature localization and recognition problem. The typical SSD is designed for 2-D image objection detection rather than 3-D feature recognition. Therefore, the network architecture and output of SSD are modified to fulfil the purpose of this research. In addition, some advanced techniques are also utilized to further enhance the recognition performance. Experimental results on the benchmark dataset confirm that the proposed method achieves the state-of-the-art feature recognition performance (95.20% F-score), localization performance (90.62% F-score), and recognition efficiency (243.85 ms per model).

Abstract In this paper, a multiple criteria decision making (MCDM) method based on weighted Archimedean power partitioned Bonferroni aggregation operators of generalised orthopair membership grades (GOMGs) is proposed. Bonferroni mean operator, geometric Bonferroni mean operator, power average operator, partitioned average operator, and Archimedean T-norm and T-conorm operations are introduced into generalised orthopair fuzzy sets to develop the Bonferroni aggregation operators. Their formal definitions are provided, and generalised and specific expressions are constructed. On the basis of the specific operators, a method for solving the MCDM problems based on GOMGs is designed. The working process, characteristics, and feasibility of the method are, respectively, demonstrated via a numerical example, a qualitative comparison at the aspect of characteristics, and a quantitative comparison using the example as benchmark. The demonstration results show that the proposed method is feasible that has desirable generality and flexibility in the aggregation of criterion values and concurrently has the capabilities to deal with the heterogeneous interrelationships of criteria, reduce the negative influence of biased criterion values, and capture the risk attitudes of decision makers.

An effective representation of additive manufacturing (AM) data is important for ensuring the repeatability of AM processes and the reproducibility of AM parts. Recently, several standardised representations have been developed and used in the industry. While at the same time, a number of other representations have been presented within the academia. The coexistence of different representations generates a series of questions and discussions: What is a representation of AM data? Are the standardised representations comprehensive enough to ensure the repeatability and reproducibility? What challenges have been addressed so far in the presented representations? What are the strengths and weaknesses of each representation? What are the main issues in the field of AM data representation currently? What are the potential research directions of AM data representation in the future? To approach these questions, a review of the existing representations of AM data is presented in this paper. Firstly, an in-depth analysis of the existing representations is provided. Then, detailed comparisons among these representations are made, and a discussion about the main issues in AM data representation is carried out on the basis of the comparisons. Finally, some future research directions of AM data representation are suggested.

The vertical distribution of plant physiological composition plays an important role in vegetation growth and carbon stock. The hyperspectral LiDAR was thought to be the most promising solution to assess the vertical distribution of vegetation physiological parameters, such as LAI (Leaf Area Index) and chlorophyll content. However, the instrument was not fully feasible, so the simulation and experiment of its response to these parameters was necessary. In this paper, the hyperspectral LiDAR waveform was simulated with the consideration of single scatter of plant radiative transfer model. The variation of LAI and chlorophyll along plant height was investigated and tree scenarios were assumed in the simulation. The hyperspectral lidar data experiment was also carried out to validate the simulation and method. The result showed that the hyperspectral LiDAR waveform could reflect the variation of plant physiological composition and facilitate the inversion algorithm design and it could play a significant role in parameter estimation and precision agriculture application in future.

Canopy structure plays an essential role in biophysical activities in forest environments. However, quantitative descriptions of a 3-D canopy structure are extremely difficult because of the complexity and heterogeneity of forest systems. Airborne laser scanning (ALS) provides an opportunity to automatically measure a 3-D canopy structure in large areas. Compared with other point cloud technologies such as the image-based Structure from Motion, the power of ALS lies in its ability to penetrate canopies and depict subordinate trees. However, such capabilities have been poorly explored so far. In this paper, the potential of ALS-based approaches in depicting a 3-D canopy structure is explored in detail through an international benchmarking of five recently developed ALS-based individual tree detection (ITD) methods. For the first time, the results of the ITD methods are evaluated for each of four crown classes, i.e., dominant, codominant, intermediate, and suppressed trees, which provides insight toward understanding the current status of depicting a 3-D canopy structure using ITD methods, particularly with respect to their performances, potential, and challenges. This benchmarking study revealed that the canopy structure plays a considerable role in the detection accuracy of ITD methods, and its influence is even greater than that of the tree species as well as the species composition in a stand. The study also reveals the importance of utilizing the point cloud data for the detection of intermediate and suppressed trees. Different from what has been reported in previous studies, point density was found to be a highly influential factor in the performance of the methods that use point cloud data. Greater efforts should be invested in the point-based or hybrid ITD approaches to model the 3-D canopy structure and to further explore the potential of high-density and multiwavelengths ALS data.

Maize is a widely planted crop in China and in other areas of the world and plays an important role in grain production. Monitoring the growth status of maize using remote sensing technology is an important component of precision agriculture and height, as a crucial growth indicator for maize, can be retrieved from light detection and ranging (LIDAR) data. However, height extraction for crops, such as maize using airborne laser scanning point clouds results in a great number of uncertainties and challenges. Here, airborne full-waveform LIDAR data were used to extract maize height. In the first step, a workflow was designed based on the Gold deconvolution algorithm combined with a basic data process technique. The method was then tested and was determined to be effective for capturing the portion of the waveform interacting with the tops of vegetation, characterized by lower amplitude stemming from the ground. Therefore, the number of second returns from point clouds was dramatically increased. During the experiment, the number of point clouds increased nearly 50% for three of the four maize plots, as compared with the original point clouds. Compared with the commonly used Gaussian fitting algorithm, the deconvolution algorithm had the advantage of extracting an accurate position for overlapping weak signals. The height percentiles indicated that the original and Gaussian decomposition derived point clouds data underestimated and deconvolution algorithm can accurately reflect the true height of maize, particularly for the 75% and 95% height percentiles.

Timely and accurate measurements of forest parameters are critical for ecosystem studies, sustainable forest resources management, monitoring and planning. This paper presents a processing chain for individual tree segmentation over large areas with airborne LiDAR 3D point cloud and very high resolution (VHR) optical imagery. The proposed processing chain consists of forest stand level delineation with optical imagery, individual tree segmentation with Canopy Height Model (CHM) derived from LiDAR point cloud, rough characterization of trees at forest stand level, and point clustering of individual tree with an Adaptive Mean Shift 3D (AMS3D) algorithm. The processing chain is developed with the expectation of supporting operational forest inventory at individual tree level. Experiment is conducted using LiDAR data acquired in Ventoux region, France. Results suggest that the proposed processing chain can be successfully adopted for individual tree characterization over large areas with different forest stands.

This letter introduces a new approach for light detection and ranging (LiDAR) waveform decomposition. First, inflection points are identified by the Ramer-Douglas-Peucker curve-fitting algorithm, and each inflection point has a corresponding baseline during curve fitting. Second, according to the spatial relation between the baseline and the inflection point, peaks are selected from the inflection points. The distance between each peak and its baseline and the maximum number of peaks are employed as a criterion to select a “significant” peak. Initial parameters such as width and boundaries of peaks provide restraints for the decomposition; right and left boundaries are estimated via a conditional search. Each peak is fitted by a Gaussian function separately, and other parts of the waveform are fitted as line segments. Experiments are implemented on waveforms acquired by both small-footprint LiDAR system LMS-Q560 and large-footprint LiDAR system Laser Vegetation Imaging Sensor. The results indicate that the algorithm could provide an optimal solution for LiDAR waveform decomposition.