3D vehicle detectors based on point clouds generally have higher detection performance than detectors based on multi-sensors. However, with the lack of texture information, point-based methods get many missing detection of occluded and distant vehicles, and false detection with high-confidence of similarly shaped objects, which is a potential threat to traffic safety. Therefore, in the long run, fusion-based methods have more potential. This paper presents a multi-level fusion network for 3D vehicle detection from point clouds and images. The fusion network includes three stages: data-level fusion of point clouds and images, feature-level fusion of voxel and Bird’s Eye View (BEV) in the point cloud branch, and feature-level fusion of point clouds and images. Besides, a novel coarse-fine detection header is proposed, which simulates the two-stage detectors, generating coarse proposals on the encoder, and refining them on the decoder. Extensive experiments show that the proposed network has better detection performance on occluded and distant vehicles, and reduces the false detection of similarly shaped objects, proving its superiority over some state-of-the-art detectors on the challenging KITTI benchmark. Ablation studies have also demonstrated the effectiveness of each designed module.

This paper aims to control the in-plane wave propagation characteristics of two-dimensional functionally graded periodic grid structures by adding local resonators made of the spring oscillator. The in-plane wave dispersion relations of the periodic grid structures with local resonators are calculated by spectral element method combined with Bloch theorem. Meanwhile, the effectiveness of the band gap calculation method is verified by the vibration transmission of the finite-length grid structures calculated based on spectral element method and finite element method. According to the spectral element method, the spectral stiffness matrix of the functionally graded beam element is established at first, then assembled with the additional stiffness matrix of the spring oscillator subsystem to obtain the complete spectral stiffness matrix of the functionally graded oscillator coupled beam. Finally, the whole stiffness matrix of grid structures with local resonators can be obtained by the coordinate transformation matrix, to form the in-plane wave dispersion relations equation based on the Bloch periodic boundary conditions. In addition, the effects of structural and material parameters on the in-plane wave propagation characteristics are analyzed, which can be applied to the vibration reduction design of periodic grid structures.

The automated harvesting of blueberries using a picking robot places a greater demand on the 3D spatial perception performance, as the robot’s grasping mechanism needs to pick blueberry fruits accurately at specific positions and in particular poses. To achieve this goal, this paper presents a method for blueberry detection, 3D spatial localization, and pose estimation using visual perception, which can be deployed on an OAK depth camera. Firstly, a blueberry and calyx scar detection dataset is constructed to train the detection network and evaluate its performance. Secondly, the blueberry and calyx scar detection model based on a lightweight YOLOv11 (the eleventh version of You Only Look Once) network with an improved depth-wise separable convolution (DSC) module is designed, and a 3D coordinate system relative to the camera is established to calculate the 3D pose of the blueberry fruits. Finally, the above detection model is deployed using the OAK depth camera, leveraging its depth estimation module and three-axis gyroscope to obtain the 3D coordinates of the blueberry fruits. The experimental results demonstrate that the method proposed in this paper can accurately identify blueberry fruits at various maturity levels, achieving a detection accuracy of 95.8% mAP50-95, a maximum positioning error of 7.2 mm within 0.5 m, and an average 3D pose error of 19.2 degrees (around 10 degrees at the ideal picking angle) while maintaining a detection frame rate of 13.4 FPS (frames per second) on the OAK depth camera, providing effective picking guidance for the mechanical arm of picking robots.

In autonomous driving, 3D vehicle detection plays an important role in traffic safety, and the detection of occluded vehicles is equally important. Different from 2D images, the point clouds are sparse and disordered, causing an uncompetitive expression for occluded vehicles. As a result, current detection methods cannot achieve a higher performance for the detection of occluded vehicles. To alleviate this problem, we present a novel 3D detector for occluded vehicle (3DOV) detection using lidar data. The proposed method has the following contributions: different data representations, including a bird’s-eye view and voxel grids, which are fused to strongly represent the features of occluded vehicles; a novel voxel-wise convolutional encoder that leverages the hierarchical property of convolutional networks to learn features with spatially local correlation from unordered points; a multiscale detection header that removes the influence of occlusions and optimizes the backpropagation of gradients; and a slimmer 3DOV network is presented to handle the speed-accuracy tradeoff. The experiment results on the KITTI 3D vehicle-detection benchmark show that the 3DOV has achieved excellent performance in terms of accuracy and efficiency, especially for occluded vehicles. Ablation studies are provided to demonstrate the effectiveness of each designed module.

3D object detection has recently become a research hotspot in the field of autonomous driving. Although great progress has been made, it still needs to be further improved. Therefore, this paper presents FDCA, a feature deep continuous aggregation network using multi-sensors for 3D vehicle detection. The proposed network adopts a two-stage structure with the bird’s-eye view (BEV) map and the RGB image as an input. In the first stage, two feature extractors were used to generate feature maps with the high-resolution and representational ability for each input view. These feature maps were then fused and fed to a 3D proposal generator to obtain the reliable 3D vehicle proposals. In the second stage, the refinement network aggregated the features of the proposal regions further and performed classifications, a 3D bounding boxes regression, and orientation estimations to predict the location and heading of vehicles in 3D space. The FDCA network proposed was trained and evaluated on the KITTI 3D object detection benchmark. The experimental results of the validation set illustrated that compared with other fusion-based methods, the 3D average precision (AP) could achieve 76.82% on a moderate setting while having real-time capability, which was higher than that of the second-best performing method by 2.38%. Meanwhile, the results of ablation experiments show that the convergence rate of FDCA was much faster and the stability was also much better, making it a candidate for application in autonomous driving.

Automated harvesting of blueberries using a picking robot places a greater demand on the 3D spatial perception performance of blueberry fruits, as the robot’s grasping mechanism needs to pick blueberry fruits accurately at specific positions and in particular poses. To achieve this goal, this paper presents a method for blueberry detection, 3D spatial localization, and pose estimation using visual perception, which can be deployed on an OAK depth camera. Firstly, a blueberry and calyx scar detection dataset is constructed to train the detection network and evaluate its performance. Secondly, the blueberry and calyx scar detection model based on a lightweight yolov11 (the eleventh version of the You Only Look Once) network with an improved depth-wise separable convolution (DSC) module is designed, and a 3D coordinate system relative to the camera is established to calculate the 3D pose of the blueberry fruits. Finally, the above detection model is deployed onto the OAK depth camera, leveraging its depth estimation module and three-axis gyroscope to obtain the 3D coordinates of the blueberry fruits. Experimental results demonstrate that the method proposed in this paper can accurately perceive blueberry fruits at various maturity levels, achieving a detection accuracy of 95.8% mAP50-95, a positioning error of less than 1 centimeter within 0.5 meters, and an average 3D pose error of 19.2 degrees while maintaining the detection frame rate of 13.4 FPS (Frames Per Second) on the OAK depth camera, which can provide effective picking guidance for the mechanical arm of picking robot.