3D LiDAR maps are playing a more and more important role in intelligent vehicle localization. In this paper, the proposed SeqPolar localization method consists of two parts. The first part is a novel node-based LiDAR map representation method, termed polarized LiDAR map (PLM), and the second is a map matching algorithm based on a second-order hidden Markov model (HMM2) for vehicle localization. Unlike existing 3D LiDAR map, which is usually constructed by accumulating 3D LiDAR clouds collected at different times, the proposed polarized LiDAR map is generated from a set of nodes. Each node consists of three elements: a polarized LiDAR image, scene features extracted from the polarized LiDAR image, and sensor pose. The polarized LiDAR image encodes 3D coordinates and reflectivities of the 3D LiDAR cloud using a multi-channel image format, a more concise, straightforward, and structured representation of the 3D LiDAR cloud. In the localization step, we propose an HMM2-based method to match a sequence of input polarized images within the map nodes and find the nearest node from the PLM. Afterward, the vehicle is readily localized from the matched map node with 3D registration. The proposed SeqPolar localization method has been validated with the actual field dataset and the public KITTI database. Experimental results demonstrate that the proposed HMM2-based matching method can achieve up to 98% accuracy to find the nearest node from PLM. Moreover, the SeqPolar localization method based on the pre-built PLM can achieve 30-centimeter localization accuracy in average on both test datasets.

Matching high definition (HD) map and real-time sensor data, which is crucial for map-based intelligent vehicle localization, usually requires robust and distinct road features extracted from both ends. However, road features extacted from sensor data (e.g., image, LiDAR) are susceptible to various conditions, such as lighting, weather conditions, dynamic objects, etc. Moreover, although lane lines can provide continuous lateral constraints on a vehicle's position, it is still challenging to get sufficient longitudinal ones due to the limited road markings. In this paper, we employed the longitudinal distance provided by road-pulse, which is the vibration pattern while the vehicle passes lateral road elements (LREs) such as speed bumps, expansion joints, etc., to enhance the localization accuracy. Such road-pulse vibration patterns can be detected accurately with the proposed Sliding Window Gaussian Model (SWGM) from the acceleration data collected using a low-cost inertial measurement unit (IMU). Detection of road-pulse allows accurate and robust matching of LREs within the HD map to provide accurate longitudinal distances. Then the longitudinal and the lateral distance(s) derived from IMU and off-the-shelf around view monitoring (AVM) are mapped into global constraints with the support of the HD map. Finally, these constraints, together with the measurement from the custom-level inertial navigation system (INS), are fused within a framework of Kalman Filter based on the Markov model (KF-MM) to achieve accurate vehicle localization. The proposed method has been verified on two routes with a total length of 3.7km. Experimental results demonstrate that road-pulse can significantly enhance the localization results with good robustness to different scenarios. After fusing with road-pulses, the overall localization error can achieve an average decimeter accuracy on both test routes.

Road traffic sign recognition is an essential function of modern autonomous driving systems. Accurate recognition plays an important role in ensuring vehicle safety and enabling subsequent operations. This paper introduces a Siamese architecture that is combined with feature reconstruction to overcome current method limitations. In detail, the proposed method enhances recognition accuracy by reconstructing convolutional features within a Siamese Neural Network (SNN) framework, integrating an improved Mamba network with mainstream CNN architectures, including VGG-16, AlexNet, ResNet, or MobileNetV2. Through comprehensive architectural comparisons, the optimal network configuration is determined for different application scenarios. Utilizing this approach, the accuracy of traffic sign recognition is substantially improved, addressing the shortcomings of existing technologies. Extensive experimental validation was conducted on multiple datasets. Taking VGG-16 network as an example, the experimental results demonstrated the method’s effectiveness, achieving accuracies of 99.83% on the GTSRB dataset, 99.13% on the TSRD dataset, and 99.07% on the TT100K dataset.

Accurate and robust self-localization is a crucial task for intelligent vehicles. Because of limited access to GPS signals, localization in underground parking lots remains a problem. In this paper, fusion localization for intelligent vehicles using the widely available around view monitoring (AVM) is conducted by Kalman filter based on second-order Markov motion model (KF-MM2). The proposed method consists of two steps, one for visual map construction from AVM images and the other for map-based multi-scale localization. The proposed visual map consists of a series of nodes. Each node encodes both holistic and local visual features computed from AVM images, three-dimensional structure, and vehicle pose. In the localization step, the process of image-level localization is modeled as a Hidden Markov Model (HMM), in which the map nodes are hidden states. The result of image-level localization is calculated using forward algorithm by the given AVM image sequence. Then the metric localization is computed from local features matching. Finally, the metric localization is fused with the prediction by KF-MM2. The proposed method has been verified in two typical underground parking lots. Experimental results demonstrate that the proposed method can achieve an average error of 0.39 m in underground parking lots.

In order to solve the problem that when installing the slit lamp in the rotary drum, the human eye has an uncertainty of calculating the center distance of the cross-hair in the drum image, a digital adjustment technique based on the improved probabilistic Hough transform was proposed. By preprocessing the original image, improving the probability Hough line detection, and formulating the line intersection screening rules, the high-precision drum image cross-hair straight line detection and cross-hair center distance calculation were realized. Experiments show that the improved probability Hough transform can accurately detect the crosshairs in the image and fit the excess straight line 100% into 4 straight lines; and the intersection point screening rule can be used to accurately screen out 2 effective intersection points. The adjustment technology can realize digital assembly of rotary drum and the digital information can be used to remind the workers of the adjustment of drums better or, which can meet the actual needs of the drum production line.

To solve the problem of mobile robots needing to adjust their pose for accurate operation after reaching the target point in the indoor environment, a localization method based on scene modeling and recognition has been designed. Firstly, the offline scene model is created by both handcrafted feature and semantic feature. Then, the scene recognition and location calculation are performed online based on the offline scene model. To improve the accuracy of recognition and location calculation, this paper proposes a method that integrates both semantic features matching and handcrafted features matching. Based on the results of scene recognition, the accurate location is obtained through metric calculation with 3D information. The experimental results show that the accuracy of scene recognition is over 90%, and the average localization error is less than 1 meter. Experimental results demonstrate that the localization has a better performance after using the proposed improved method.