Category-level 6D object pose estimation aims to predict the position and orientation of unseen object instances, which is a fundamental problem in robotic applications . Previous works mainly focused on exploiting visual cues from RGB images , while depth images received less attention. However, depth images contain rich geometric attributes about the object’s shape, which are crucial for inferring the object’s pose. This work achieves category-level 6D object pose estimation by performing sufficient geometric learning from depth images represented by point clouds. Specifically, we present a novel geometric consistency and geometric discrepancy learning framework called CD-Pose to resolve the intra-category variation, inter-category similarity, and objects with complex structures. Our network consists of a Pose-Consistent Module and a Pose-Discrepant Module. First, a simple MLP-based Pose-Consistent Module is utilized to extract geometrically consistent pose features of objects from the pre-computed object shape priors for each category. Then, the Pose-Discrepant Module, designed as a multi-scale region-guided transformer network, is dedicated to exploring each instance’s geometrically discrepant features. Next, the NOCS model of the object is reconstructed according to the integration of consistent and discrepant geometric representations . Finally, 6D object poses are obtained by solving the similarity transformation between the reconstruction and the observed point cloud. Experiments on the benchmark datasets show that our CD-Pose produces superior results to state-of-the-art competitors.

Category-level object pose estimation aims to predict the 6D pose and 3D metric size of objects from given categories. Due to significant intra-class shape variations among different instances, existing methods have mainly focused on estimating dense correspondences between observed point clouds and their canonical representations, i.e., normalized object coordinate space (NOCS). Subsequently, a similarity transformation is applied to recover the object pose and size. Despite these efforts, current approaches still cannot fully exploit the intrinsic geometric features to individual instances, thus limiting their ability to handle objects with complex structures (i.e., cameras). To overcome this issue, this paper introduces GPT-COPE, which leverages a graph-guided point transformer to explore distinctive geometric features from the observed point cloud. Specifically, our GPT-COPE employs a Graph-Guided Attention Encoder to extract multiscale geometric features in a local-to-global manner and utilizes an Iterative Non-Parametric Decoder to aggregate the multiscale geometric features from finer scales to coarser scales without learnable parameters. After obtaining the aggregated geometric features, the object NOCS coordinates and shape are regressed through the shape prior adaptation mechanism, and the object pose and size are obtained using the Umeyama algorithm. The multiscale network design enables perceiving the overall shape and structural information of the object, which is beneficial to handle objects with complex structures. Experimental results on the NOCS-REAL and NOCS-CAMERA datasets demonstrate that our GPT-COPE achieves state-of-the-art performance and significantly outperforms existing methods. Furthermore, our GPT-COPE shows superior generalization ability compared to existing methods on the large-scale in-the-wild dataset Wild6D and achieves better performance on the REDWOOD75 dataset, which involves objects with unconstrained orientations.

The lunar surface temperature (LST) is one of the important thermophysical parameters of the Moon, which helps to study the radiative properties of the lunar surface. Thermal infrared emission spectra are sensitive to the thermophysical properties of the lunar surface material, and the emissivity data can be used for lunar surface composition inversion. In the past, humans have conducted hundreds of lunar exploration missions, but only a small number have been conducted in the infrared band, and typical examples include the Apollo program and the Diviner lunar radiometer experiment of the LRO satellite. Only part of the lunar surface has been explored by these missions. Sustainable Development Goals Satellite-1 (SDGSAT-1) carries out a complete observation of the lunar surface in three infrared wavelength bands (B1: 8–$10.5~\mu \text{m}$, B2: 10.3–$11.3~\mu \text{m}$, and B3: 11.5–$12.5~\mu \text{m}$) by its thermal imager spectrometer as part of its mission. In this study, the temperature-emissivity separation (TES) algorithm is used to retrieve the LST and calculate the thermal infrared band emissivity using the radiometric data obtained by SDGSAT-1 thermal imager spectrometer, and the temperature distribution of full-disk Moon and the emissivity distribution of three bands are also mapped. The temperature retrieval error is verified less than 1 K.

This paper presents 6D vision transformer (6D-ViT), a transformer-based instance representation learning network suitable for highly accurate category-level object pose estimation based on RGB-D images. Specifically, a novel two-stream encoder-decoder framework is dedicated to exploring complex and powerful instance representations from RGB images, point clouds, and categorical shape priors. The whole framework consists of two main branches, named Pixelformer and Pointformer. Pixelformer contains a pyramid transformer encoder with an all-multilayer perceptron (MLP) decoder to extract pixelwise appearance representations from RGB images, while Pointformer relies on a cascaded transformer encoder and an all-MLP decoder to acquire the pointwise geometric characteristics from point clouds. Then, dense instance representations (i.e., correspondence matrix and deformation field) for NOCS model reconstruction are obtained from a multisource aggregation (MSA) network with shape prior, appearance and geometric information as inputs. Finally, the instance 6D pose is computed by solving the similarity transformation between the observed point clouds and the reconstructed NOCS representations. Extensive experiments with synthetic and real-world datasets demonstrate that the proposed framework achieves state-of-the-art performance for both datasets. Code is available at https://github.com/luzzou/6D-ViT.

The public transport information has been focus of social attention, especially bus arrival time (BAT) prediction. Historical data in combination with real-time data may be used to predict the future travel times of vehicles more accurately, thus improving the experience of the users who rely on such information. In this paper, we expound the correspondence among real-time data, history data and BAT. Hence, we propose short distance BAT prediction based on real-time traffic condition and long distance BAT prediction based on K Nearest Neighbors(KNN) respectively. Furthermore, original matching algorithm of KNN is modified for two times to accelerate matching procedure in terms of computationally expensive queries. In empirical studies with real data from buses, the model in this paper outperforms ANN or KNN used alone both in accuracy and efficiency of the algorithm, errors of which is less than 12 percent for a time horizon of 60 minutes.

Modern avionics architecture is evolving from traditional federated architecture to Integrated Modular Avionics (IMA) architecture. ARINC653 standard, which is employed in the avionics industry, supports partitioning core concept in IMA. Furthermore, avionic software has very high safety and reliability requirements in safety- critical domains. Therefore, how to develop high-integrity avionics software constructed on ARINC653 architecture becomes a very significant problem nowadays. In this paper we propose an automatic RT-Java code generation approach based on the AADL model for ARINC653 (AADL653) to enable the development of RT-Java ARINC653-based avionics software more productive and trustworthy. Our main contribution in this paper includes: (1) a mapping from the AADL653 model to a high-integrity RT-Java programming model for ARINC653 (RT-Java653); (2) an ARINC653-compliant RT-Java code generation algorithm suitable for complex multi-task collaboration interaction situation. Accordingly, we implement this RT-Java class library and corresponding code generator. Moreover, a simplified multi-task flight application as a case study is given to illustrate our approach and the preliminary experiment results show the validity of our approach.

Avionics software is safe-critical embedded software and its architecture is evolving from traditional federated architectures to Integrated Modular Avionics (IMA) to improve resource usability. ARINC653, as a standard widely employed in the avionics industry, supports partitioning concepts in accordance with the IMA philosophy. To insure the development of the avionics software constructed on ARINC653 operating system with high reliability and efficiency, we propose a model-driven design methodology based on Architecture Analysis &Design Language (AADL) for ARINC653 system. This paper focus on the modeling parts of this methodology which main feature is separating the abstract application function logic represented by AADL Platform-Independent Model (AADL PIM) from the concrete execution architecture represented by AADL Model for ARINC653 (AADL653). Additionally, we provide a refined transformation framework with formally transformation rules to transform AADL PIM to AADL653 automatically and the transformation result model AADL653 can then be used for analysis, verification and code generation.