In maintaining roads and ensuring safety, promptly detecting and repairing pavement defects is crucial. However, conventional detection methods demand substantial manpower, incur high costs, and suffer from low efficiency. To enhance road maintenance efficiency and reduce costs, we propose an improved algorithm based on YOLOv8. Our method incorporates several key enhancements. First, we replace conventional convolutions with a module composed of spatial-to-depth layers and nonstrided convolution layers (SPD-Conv) in the network backbone, enhancing the capability of recognizing small-sized defects. Second, we replace the neck of YOLOv8 with the neck of the ASF-YOLO network to fully integrate spatial and scale features, improving multiscale feature extraction capability. Additionally, we introduce the FasterNet block from the FasterNet network into C2f to minimize redundant computations. Furthermore, we utilize Wise-IoU (WIoU) to optimize the model’s loss function, which accounts for the quality factors of objects more effectively, enabling adaptive learning adjustments based on samples of varying qualities. Our model was evaluated on the RDD2022 road damage dataset, demonstrating significant improvements over the baseline model. Specifically, with a 2.8% improvement in $mAP$ and a detection speed reaching 43 FPS, our method proves to be highly effective in real-time road damage detection tasks.

The visual-inertial simultaneous localization and mapping (SLAM) is a feasible indoor positioning system that combines the visual SLAM with inertial navigation. There are accumulated drift errors in inertial navigation due to the state propagation and the bias of the inertial measurement unit (IMU) sensor. The visual-inertial SLAM can correct the drift errors via loop detection and local pose optimization. However, if the trajectory is not a closed loop, the drift error might not be significantly reduced. This paper presents a novel pedestrian dead reckoning (PDR)-aided visual-inertial SLAM, taking advantage of the enhanced vanishing point (VP) observation. The VP is integrated into the visual-inertial SLAM as an external observation without drift error to correct the system drift error. Additionally, the estimated trajectory’s scale is affected by the IMU measurement errors in visual-inertial SLAM. Pedestrian dead reckoning (PDR) velocity is employed to constrain the double integration result of acceleration measurement from the IMU. Furthermore, to enhance the proposed system’s robustness and the positioning accuracy, the local optimization based on the sliding window and the global optimization based on the segmentation window are proposed. A series of experiments are conducted using the public ADVIO dataset and a self-collected dataset to compare the proposed system with the visual-inertial SLAM. Finally, the results demonstrate that the proposed optimization method can effectively correct the accumulated drift error in the proposed visual-inertial SLAM system.

Modern inertial navigation system (INS) equipped with high-precision inertial measurement units (IMUs) estimates the pose in real-time with centimeter-level accuracy. However, its drift errors accumulate over a long trajectory because of the bias of IMUs. Therefore, to correct the drift errors and improve the localization accuracy of the system, this paper proposes a novel visual-aided INS for use in intelligent vehicles operating in artificial indoor environments. The vanishing point (VP) extracted from images is used as an external observation without drift errors in the proposed system to estimate the bias of IMUs in an optimization window. Particularly, the sliding window-based local optimization method and the segmentation window-based global optimization method are proposed to improve the system’s localization performance in accuracy under the assumption that the bias of IMUs in an optimization window is a random constant. Considering the effect of scale drifts, the proposed system is scaled using odometry. To evaluate the system’s performance, a series of experiments are conducted in underground parking lots. The experimental results demonstrate that compared to the INS, the average localization errors of the VP-aided INS are improved by 30.77% (0.18m) and 66.99% (0.34m) for Vehicle_01 and Vehicle_02, respectively. And its average attitude errors are improved by 36.36% (0.14°) and 66.26% (0.28°) for Vehicle_01 and Vehicle_02, respectively. Additionally, the proposed optimization methods also apply to the mobile platform equipped with low-cost IMUs, where the localization errors can be reduced by at least 30.46%.

Aiming at the problems of insufficient autonomous obstacle avoidance performance of UAVs in complex indoor environments, an improved Double DQN algorithm based on deep reinforcement learning is proposed. The algorithm enhances the perception and learning capabilities by optimizing the network model and employs a dynamic exploration strategy that encourages exploration in the early stage and reduces it later to accelerate convergence and improve efficiency. Simulation experiments in two scenarios of varying complexity, using an indoor simulation environment built with AirSim and UE4(Unreal Engine 4), show that in the simpler scenario, the average cumulative reward increased by 22.88%, the maximum reward increased by 101.56%, the average safe flight distance increased by 23.17%, and the maximum safe flight distance by 105.62%. In the more complex scenario, the average cumulative reward increased by 2.66%, the maximum reward increased by 88.77%, the average safe flight distance increased by 2.05%, and the maximum safe flight distance by 84.68%.

As a classic framework for visual object tracking, the Siamese convolutional neural network has received widespread attention from the research community. This method uses a convolutional neural network to obtain the object features and to match them with the search area features to achieve object tracking. In this work, we observe that the contribution of each convolution kernel in the convolutional neural network for object tracking tasks is different. We propose an object-aware convolution kernel attention mechanism. Based on the characteristics of each object, the convolution kernel features are dynamically weighted to improve the expression ability of object features. The experiments performed using OTB and VOT benchmark datasets show that the performance of the tracking method fused with the convolution kernel attention mechanism is significantly better compared with the original method. Moreover, the attention mechanism can also be integrated with other tracking frameworks as an independent module to improve the performance.

This article presents a numerically stable algorithm for analytic inverse kinematics of 7-DoF S-R-S manipulators with joint limit avoidance. The arm angle is used to represent the self-motion manifold within a global arm configuration. The joint limits are analytically mapped to the arm angle space for joint limit avoidance. To profile the relation between the joint angle and arm angle, it is critical to characterize the singular arm angle for each joint. In the-state-of-the art methods, the existence of the singular arm angle is triggered by comparing a discriminant with zero given a threshold. This leads to numerical issues since the threshold is inconsistent among different target poses, leading to incorrect range of the arm angle. These issues are overcome by associating indeterminate joint angles of tangent joints with angles of 0 or ±π of cosine joints, rather than using an independent threshold for each joint. The closed-form algorithm in c++ code to perform numerically stable inverse kinematics of 7-DoF S-R-S manipulators with global arm configuration control and joint limit avoidance is also given.