9 June 2025 in IEEE Transactions on Instrumentation and Measurement
Superficial defect detection on bridge piers is crucial for assessing the health condition of bridges. A novel ring vision acquisition system based on a climbing operation robot is proposed to aim at the limitations of existing Unmanned Aerial Vehicles (UAVs) and wall-climbing robots. In addition, to address the quadratic computational complexity of the Transformer-based model, a Vision Mamba-based defects detection model, Piers Defects-Mamba (PD-Mamba), is proposed, which contributes to more efficient inferencing the high-resolution defects images. Finally, field experiments were performed at Shouchun Bridge in Anhui Province, China. The results demonstrated that the robotic system could automate acquiring high-precision visual images of bridge piers’ superficial defects. The proposed PD-Mamba achieves 77.97% mean Intersection over Union (mIoU), 88.81% mean Precision (mP), 85.42% mean Recall (mR), and 87.05% mean F1-score (mF1) on the multi-category bridge pier defect dataset, exhibiting state-of-the-art (SOTA) performance, and has global modelling capability and linear computational complexity, which achieves accurate and efficient detection of bridge piers’ superficial defects. It is important to assess the health condition of the bridge.
11 December 2025 in IEEE Transactions on Intelligent Transportation Systems
Vehicle visual perception is essential for safe autonomous driving, especially in challenging low-visibility conditions. Polarimetric imaging has shown enhanced perception by improving target-background contrast and reducing glare. However, current research in polarimetric imaging for autonomous driving largely focuses on singular use of polarimetric features. Exploiting polarimetric information to enhance vehicle visual perception in low-visibility scenarios remains a critical challenge. This paper addresses this challenge by integrating multiple polarimetric parameter characterization with a deep learning model for semantic segmentation visual perception task, which is called TransWNet. The model combines Convolutional Neural Networks (CNN) and Transformer architectures, extracting features and contextual information of both Degree of Polarization (DoP) and Angle of Polarization (AoP) comprehensively during the encoding phase. In the decoding phase, it incorporates skip connections to effectively retain multimodal polarimetric information across deep and shallow features, generating output through feature fusion. To the best of our knowledge, this is the first work that jointly exploits DoP and AoP for vehicle scene understanding in degraded visibility. Experimental results demonstrate that TransWNet, by effectively leveraging multimodal polarimetric information, achieves significantly better performance in semantic segmentation of low-visibility traffic scenes, with marked improvements in mIoU, mPA, and Accuracy over all single-feature baselines. Compared with the baseline method, TransWNet improves the Accuracy by 3.44%.
21 July 2026 in Journal of Field Robotics