SCRnet: A Spatial Consistency Guided Network Using Contrastive Learning for Point Cloud Registration

Point cloud registration is used to find a rigid transformation from the source point cloud to the target point cloud. The main challenge in the point cloud registration is in finding correct correspondences in complex scenes that may contain many noise and repetitive structures. At present, many existing methods use outlier rejections to help the network obtain more accurate correspondences, but they often ignore the spatial consistency between keypoints. Therefore, to address this issue, we propose a spatial consistency guided network using contrastive learning for point cloud registration (SCRnet), in which its overall stage is symmetrical. SCRnet consists of four blocks, namely feature extraction block, confidence estimation block, contrastive learning block and registration block. Firstly, we use mini-PointNet to extract coarse local and global features. Secondly, we propose confidence estimation block, which formulate outlier rejection as confidence estimation problem of keypoint correspondences. In addition, the local spatial features are encoded into the confidence estimation block, which makes the correspondence possess local spatial consistency. Moreover, we propose contrastive learning block by constructing positive point pairs and hard negative point pairs and using Point-Pair-INfoNCE contrastive loss, which can further remove hard outliers through global spatial consistency. Finally, the proposed registration block selects a set of matching points with high spatial consistency and uses these matching sets to calculate multiple transformations, then the best transformation can be identified by initial alignment and Iterative Closest Point (ICP) algorithm. Extensive experiments are conducted on KITTI and nuScenes dataset, which demonstrate the high accuracy and strong robustness of SCRnet on point cloud registration task.

Climate change in the Western Balkans and EU Green Deal: status, mitigation and challenges

Background The European Commission (EC), based on the European Green Deal (2019) and the Recovery plan for Europe (2021), envisages investing 30% of the budget in climate-related programs, projects, and initiatives, which clearly shows Europe's commitment to becoming the first climate-neutral region by 2050. Activities are also planned for countries that are not members of the European Union (EU), which require complex changes in the field of legislation, strategic planning, implementation, and monitoring. To successfully plan short-term and long-term activities on these grounds, it is necessary to have a realistic picture of the state of climate change in each country—as they spill over into the entire region of Europe. The main objective of this paper is to answer the following questions: (i) is climate change observed in Western Balkans? (ii) how are certain sectors vulnerable to climate change in Western Balkans? (iii) what are the climate change adaptation strategies in the six countries of the Western Balkans? The answers to these questions can help in planning activities and initial alignment of Western Balkan countries with the EU plan to achieve net zero greenhouse gas (GHG) emissions by 2050. Main body The main results of the research show that in all countries of the region, the average annual temperature increased by 1.2 °C compared to 1970, with stabilization and the beginning of the decline which can be expected around 2040. The main reasons for climate change in the region are: industry, energy, and heating sector based on coal exploitation, low energy efficiency, etc. Conclusions It can be concluded that Croatia as a member of the EU has adopted, and other five Western Balkans countries are in the process of adopting the necessary regulations and strategies towards climate change mitigation, but the implementation of specific activities is at a low level. The reason for this most often lies in the insufficient commitment of decision-makers to make significant changes in the field of climate change transition (lower level of economic development, lack of investment, and preservation of social peace). Finally, this paper provides an overview of climate change by country, scenario analysis, and policy recommendations.

Neutron Reflectometry-Based In Situ Structural Analysis of an Aligning Agent Additive for the Alignment of Nematic Liquid Crystals on Solid Substrates

Surface aligning agents, such as amphiphilic surfactants, are widely used to control the initial alignment of nematic liquid crystals (NLCs) in liquid crystal displays (LCDs). Generally, these agents are first...

On SINS\Star Tracker Geo-localization

This work focuses on the feasibility of a fully autonomous geo-localization system for near-earth applications based on the strap-down inertial navigation system (SINS) and the star tracker. First, each sensor is analyzed individually. Then, the performance of the integrated system in a dynamic situation is investigated. Moreover, a detailed angle error analysis is given to estimate the impact on geo-localization. The navigation solution is proven to be affected by the sensors' errors plus an algorithmic error from the dead reckoning computation. Lastly, simulations are concluded to assess the dynamic movement scenario's performance and navigational possibility using the nonlinear Kalman filter. The results show the continuing divergence of the integrated navigation system affected by the dead reckoning algorithm. However, the continuous initial alignment in static mode reinitializes the position error successfully.

In-Motion Initial Alignment Method for LDV-Aided SINS Based on Adaptive Unscented Quaternion H-infinite Filter

With the advantages of high velocity measurement accuracy and fast dynamic response, the laser Doppler velocimeter (LDV) is expected to replace the odometer (OD) to be combined with a strapdown inertial navigation system (SINS) to form a higher precision integrated navigation system. Since LDV has higher velocity measurement accuracy and data update frequency than OD and Doppler velocity log, LDV is used for the first time to aid SINS in in-motion alignment in this paper. Considering that some approximation used in the alignment model, the uncertainty noise of the sensors during the motion process and the unknown noise parameters during the filter process, an adaptive unscented quaternion H-infinite estimator (AUSQUHE) is proposed in this paper. The proposed AUSQUHE method has high robustness since it combines the advantages of unscented quaternion estimator and H-infinite filter. And the adaptive threshold of H-infinite filter and the adaptive measurement noise covariance matrix are introduced to make the filter adapt to the environment change and accelerate the convergence of errors. The performance of the proposed method is verified by a vehicle field test with normal LDV signal and a vehicle test with LDV signal disturbed by the noise. The results show that the proposed method has higher alignment accuracy, faster convergence speed and stronger robustness than four other compared methods.