Place recognition is a fundamental task in robotics, enabling loop closure detection in simultaneous localization and mapping (SLAM), and re-localization on prior maps. Current range image-based networks use single-column convolution to maintain feature invariance to shifts in image columns caused by light detection and ranging (LiDAR) viewpoint change. However, this raises the issues such as “restricted receptive fields” and “excessive focus on local regions”, degrading the performance of networks, especially in scenarios with movable objects. In this paper, a lightweight circular convolutional Transformer network named CCTNet is proposed, which aims to boost performance by capturing structural information in point clouds and facilitating cross-dimensional interaction of spatial and channel information. Initially, a Circular Convolution Module (CCM) is introduced, expanding the network’s perceptual field while maintaining feature consistency across varying LiDAR perspectives. Then, a Range Transformer Module (RTM) is proposed, which enhances place recognition accuracy in scenarios with movable objects by employing a combination of channel and spatial attention mechanisms. Furthermore, we propose an overlap-based loss function, transforming the place recognition task from a binary loop closure classification into a regression problem linked to the overlap between LiDAR frames. Through extensive experiments on the Karlsruhe Institute of Technology and Toyota Technological Institute at Chicago (KITTI) and Ford Campus datasets, CCTNet surpasses comparable methods, achieving Recall@1 of 0.924 and 0.965, and Recall@1% of 0.990 and 0.993 on the test set, showcasing superior performance. Results on the self-collected dataset further demonstrate the proposed method’s potential for practical implementation in complex scenarios to handle movable objects, showing improved generalization across various datasets.

In this article, an inexpensive beam homogenizer is designed to transform the laser beam with irregular intensity into a uniform square beam with a large size, thus eliminating the background noise caused by uneven heating and improving the defect detection capability of the laser infrared thermography (IRT) detection method. A finite difference method simulation model for IRT detection is established, which can be combined with the optical simulation software Zemax OpticalStudio to realize the optical-thermal coupling simulation, represent the physical phenomenon more realistically, and guide the design and optimization of beam homogenizer. A unipolar square wave excitation delayed sampling thermal signal reconstruction nondestructive testing method and the corresponding image processing method are proposed, which can effectively solve the interference caused by the reflected laser on the IRT detection and weaken the problem of low contrast between defective and normal regions caused by the concentration of heat in the surface layer of the specimen during the heating process. The above-mentioned method has achieved good results in metal additive manufacturing parts that lack fusion defect detection experiments. The experimental results show that, compared with traditional methods, the novel method proposed in this article requires the collection and processing of fewer infrared images, yet achieves a superior defect detection rate and signal-to-noise ratio.

Constrained by Abbe diffraction, terahertz time-domain spectroscopy utilizing photoconductive antennas (PCA) is limited to submillimeter spatial resolution and requires several hours to complete a scan. Existing solutions involve using high-precision probes or solid immersion lenses to enable super-resolution imaging. However, these techniques necessitate precise experimental setups and controlled environments, rendering them unsuitable for meeting requirements of field use in areas, such as nondestructive testing, biomedicine, and nanotechnology. Here, the deep learning-based technique THz-super-resolution generative adversarial network (THz-SRGAN) was introduced in the THz-SR imaging field for the first time. The new imaging method not only overcomes the Abbe diffraction limit at minimal cost but also extracts valuable information regarding the physical properties of imaged objects within a narrow spatial field. Furthermore, a Richardson–Lucy algorithm was developed for THz-SR imaging and compared the performance with that of THz-SRGAN. The experimental results demonstrate that the proposed THz-SRGAN method achieves the most significant improvement in spatial resolution to-date.

Global localization is a challenging task for intelligent robots, as its accuracy directly contributes to the performance of downstream navigation and planning tasks. However, existing literature focus more on the place retrieval and the success rate of localization, with limited attention given to the metrics of position estimation. In this paper, a single-shot global LiDAR localization method is proposed with the ultimate goal of achieving high position accuracy, inspired by the positioning approach of multi-constellation localization systems. Initially, we perform coarse localization using global descriptors and select observation points along with their corresponding coordinates based on the obtained coarse localization results. Coordinates can be acquired from a pre-built map, GNSS, or other devices. Then, a lightweight LiDAR odometry method is designed to estimate the distance between the retrieved data and the observation points. Ultimately, the localization problem is transformed into an optimization problem of solving a system of multiple sphere equations. The experimental results on the KITTI dataset and the self-collected dataset demonstrate that our method achieves an average localization error (including errors in the z-axis) of 0.89 meters. In addition, it achieves retrieval efficiency of 0.357 s per frame on the former dataset and 0.214 s per frame on the latter one. Code and data are available at https://github.com/jlurobot/multi-constellation-localization.

In nighttime driving scenes, due to insufficient and uneven lighting, and the scarcity of high-quality datasets, the miss rate of nighttime pedestrian detection (PD) is much higher than that of daytime. Vision-based distance detection (DD) has the advantages of low cost and good interpretability, but the existing methods have low precision, poor robustness, and the DD is mostly performed independently of PD. A narrowband near-infrared (NIR) camera and an NIR lamp to relieve the impact of noisy visible light (VIS) and improve imaging quality have been selected; a LiDAR to obtain distance information has been embedded in the system; a nighttime driving scene pedestrian and its distance joint detection dataset NIRPed has been built. The NIRPed includes 146k pedestrian annotations, which is 3 times that of NightOwls, the largest VIS nighttime pedestrian dataset. Based on Faster-RCNN, a joint PD and DD method using monocular imaging has been proposed. The propo- sed method realizes the joint detection of pedestrian and its distance on the NIRPed dataset with the PD log-average miss rate and DD mean absolute error rate of 6.5% and 5.5%, respectively. For comparison, the joint detection method was implemented on other large-scale VIS pedestrian datasets as well. Moreover, compared with the existing vision-based DD methods, the proposed method is less affected by pedestrian distance and height, exhibits higher accuracy and robustness, and satisfied the ISO requirements for intelligent transportation systems (absolute error rate < 15%).

In this paper, terahertz time-domain spectroscopy (THz-TDS) is used for the first time to detect fabricated defects in a glass fiber-skinned lightweight honeycomb composite panel. A novel amplitude polynomial regression (APR) algorithm is proposed as a preprocessing method. This method segments the amplitude-frequency curves to simulate the heating and the cooling monotonic behavior as in infrared thermography. Then, the method of empirical orthogonal function (EOF) imaging is applied on the APR preprocessed data as a postprocessing algorithm. Signal-to-noise ratio analysis is performed to verify the image improvement of the proposed APR-EOF modality from a quantitative point of view. Finally, the experimental results and the physical analysis show that THz is more suitable with respect to the detection of defects in glass fiber lightweight honeycomb composites.

As an emerging nondestructive diagnostic and monitoring technique, terahertz time-domain spectroscopy (THz-TDS) imagery is attracting more attention. In this regard, new THz image processing algorithms based on infrared thermography (IRT) concepts are greatly needed, since most IRT imagery modalities are fast for in-line industrial inspection. However, this scenario is difficult due to some physical constraints to be reached, although this idea should be followed to avoid the loss of useful information during image processing. In this paper, a novel THz amplitude polynomial principle component regression (APPCR) algorithm is proposed for the inspection of aramid-basalt hybrid composite laminates. This algorithm segments THz amplitude-frequency curves to simulate heating-up and cooling-down behaviors as in IRT; in addition, it uses an empirical orthogonal functions-based principle component regression modality to simplify the THz image analysis procedure. This experimental and analytical study shows that APPCR can, first, simplify the THz image analysis procedure, and second, enhance image contrast and spatial resolution. A theoretical analysis was conducted as experimental explanation, while the IRT imagery results were used for comparative purposes. In addition, signal-to-noise ratio analysis was used to evaluate quantitatively the image enhancement. Finally, it is possible to conclude that THz is more suitable to inspect transparent or semitransparent materials. Advantages and disadvantages of THz-TDS and IRT are summarized in the text.

In this paper, optical and mechanical excitation thermography was used to investigate basalt-fiber-reinforced polymer, carbon-fiber-reinforced polymer, and basalt-carbon fiber hybrid specimens subjected to impact loading. Interestingly, two different hybrid structures including sandwich-like and intercalated stacking sequence were used. Pulsed phase thermography, principal component thermography, and partial least-squares thermography (PLST) were used to process the thermographic data. X-ray computed tomography was used for validation. In addition, signal-to-noise ratio analysis was used as a means of quantitatively comparing the thermographic results. Of particular interest, the depth information linked to Loadings in PLST was estimated for the first time. Finally, a reference was provided for taking advantage of different hybrids in view of special industrial applications.