Methodology and hardware for assessing the quantitative characteristics of pet images in the study of dynamic objects

The description of the original phantom design for assessing the quantitative characteristics of PET images in the study of dynamic objects is given. The phantom movement is controlled by the breath synchronization system, which records the phantom movement amplitude and the duration of the movement cycle. A curve was obtained that simulates human breathing, the parameters of which (amplitude and period) correspond to those obtained in the study of the chest. The values of the ecovery coefficients and contrast are obtained taking into account the sizes of the spheres, as well as the static and dynamic types of movement of phantoms. An assessment of the discrepancy between the recovery coefficients and the contrast values for the spheres installed inside the phantom in the static and dynamic states has been made. With a decrease in the diameter (respectively, and volume) of the sphere, an increase in the difference in values (between the static and dynamic positions of the phantom) of the recovery coefficient is observed. The optimal values of the recovery coefficients obtained using the QClear reconstruction algorithm have been determined. Recommendations for the use of the developed device in the study of dynamic objects are described. It is advisable to use the installation presented in this work to control the quality of the qualitative and quantitative characteristics of diagnostic images obtained both on PET/CT scanners and during studies using SPECT/CT (single-photon emission tomograph combined with a computed tomograph).

LIO-CSI: LiDAR inertial odometry with loop closure combined with semantic information

Accurate and reliable state estimation and mapping are the foundation of most autonomous driving systems. In recent years, researchers have focused on pose estimation through geometric feature matching. However, most of the works in the literature assume a static scenario. Moreover, a registration based on a geometric feature is vulnerable to the interference of a dynamic object, resulting in a decline of accuracy. With the development of a deep semantic segmentation network, we can conveniently obtain the semantic information from the point cloud in addition to geometric information. Semantic features can be used as an accessory to geometric features that can improve the performance of odometry and loop closure detection. In a more realistic environment, semantic information can filter out dynamic objects in the data, such as pedestrians and vehicles, which lead to information redundancy in generated map and map-based localization failure. In this paper, we propose a method called LiDAR inertial odometry (LIO) with loop closure combined with semantic information (LIO-CSI), which integrates semantic information to facilitate the front-end process as well as loop closure detection. First, we made a local optimization on the semantic labels provided by the Sparse Point-Voxel Neural Architecture Search (SPVNAS) network. The optimized semantic information is combined into the front-end process of tightly-coupled light detection and ranging (LiDAR) inertial odometry via smoothing and mapping (LIO-SAM), which allows us to filter dynamic objects and improve the accuracy of the point cloud registration. Then, we proposed a semantic assisted scan-context method to improve the accuracy and robustness of loop closure detection. The experiments were conducted on an extensively used dataset KITTI and a self-collected dataset on the Jilin University (JLU) campus. The experimental results demonstrate that our method is better than the purely geometric method, especially in dynamic scenarios, and it has a good generalization ability.

Inverse problems of the dynamics of observation interpretation systems

Based on the analysis and systematization of the inverse problems of dynamics, the study of the properties and features of the types of dynamic models under consideration, an approach is proposed for the development of appropriate methods of mathematical modeling based on the use and implementation of integral models in the form of Volterra equations of the I and II kind, their functional capabilities are determined in the study of various classes of problems, and also formulated the features that affect the choice of methods for their numerical solution. Methods for obtaining integral models are proposed, which are the basis for constructing algorithms for solving inverse problems of dynamics for a fairly wide class of dynamic objects. Integral methods for the identification of dynamic objects have been developed, which make it possible to obtain stable non-optimization algorithms for calculating the parameters of mathematical models. Recurrent methods of parametric identification of transfer functions of dynamic objects with an arbitrary input action are proposed (the obtained parameters of the transfer functions are also coefficients of the corresponding differential equations, which makes it possible to obtain equivalent mathematical models in the form of integral equations). The study of algorithms that implement the proposed identification methods allows us to conclude about their efficiency in terms of the amount of computation and ease of implementation, as well as the high accuracy of calculating the model parameters.

Identification of UAV model parameters from flight and computer experiment data

The object of research in the article is various well-known approaches and methods of structural and parametric identification of dynamic controlled objects - unmanned aerial vehicles (UAVs). The subject of the research is the parameters of linear and nonlinear mathematical models of spatial and isolated movements, describing the dynamics and aerodynamic properties of the UAV and obtained both from the results of flight experiments and using computer object-oriented programs for 3-D UAV models. The goal is to obtain mathematical models of UAV flight dynamics in the form of differential equations or transfer functions, check them for reliability and the possibility of using them in problems of synthesis of algorithms for automatic control systems of UAVs. Tasks to be solved: evaluation of the analytical (parametric), direct (transient), as well as the identification method using the 3-D model of the control object. Methods used structural and parametric identification of dynamic objects; the determination of static and dynamic characteristics of mathematical models by the type of their transient process; the System Identification Toolbox package of the MatLab environment, the Flow Simulation subsystem of the SolidWorks software and the X-Plane software environment. The experimental parameters of UAV flights, as well as the results of modeling in three-dimensional environments, are the initial data for the identification of mathematical models. The following results were obtained: the possibility of analytical and computer identification of mathematical models by highly noisy parameters of the UAV flight was shown; the mathematical models of UAVs obtained after identification is reliable and adequately reproduce the dynamics of a real object. A comparative analysis of the considered UAV identification methods is conducted, their performance and efficiency are confirmed. Conclusions. The scientific novelty of the result obtained is as follows: good convergence, reliability and the possibility of using the considered identification methods for obtaining mathematical models of dynamic objects to synthesize algorithms for automatic control systems of UAVs is shown.

Improving RGB-D SLAM in dynamic environments using semantic aided segmentation

Summary Most conventional simultaneous localization and mapping (SLAM) approaches assume the working environment to be static. In a highly dynamic environment, this assumption divulges the impediments of a SLAM algorithm that lack modules that distinctively attend to dynamic objects despite the inclusion of optimization techniques. This work exploits such environments and reduces the effects of dynamic objects in a SLAM algorithm by separating features belonging to dynamic objects and static background using a generated binary mask image. While the features belonging to the static region are used for performing SLAM, the features belonging to non-static segments are reused instead of being eliminated. The approach employs deep neural network or DNN-based object detection module to obtain bounding boxes and then generates a lower resolution binary mask image using depth-first search algorithm over the detected semantics, characterizing the segmentation of the foreground from the static background. In addition, the features belonging to dynamic objects are tracked into consecutive frames to obtain better masking consistency. The proposed approach is tested on both publicly available dataset as well as self-collected dataset, which includes both indoor and outdoor environments. The experimental results show that the removal of features belonging to dynamic objects for a SLAM algorithm can significantly improve the overall output in a dynamic scene.