Explore the vast range of titles available.

With the steady progress of China’s agricultural modernization, the demand for agricultural machinery for production is widely growing. Agricultural unmanned aerial vehicles (UAVs) are becoming a new force in the field of precision agricultural aviation in China. In those agricultural areas where ground-based machinery have difficulties in executing farming operations, agricultural UAVs have shown obvious advantages. With the development of precision agricultural aviation technology, one of the inevitable trends is to realize autonomous identification of obstacles and real-time obstacle avoidance (OA) for agricultural UAVs. However, the complex farmland environment and changing obstacles both increase the complexity of OA research. The objective of this paper is to introduce the development of agricultural UAV OA technology in China. It classifies the farmland obstacles in two ways and puts forward the OA zones and related avoidance tactics, which helps to improve the safety of aviation operations. This paper presents a comparative analysis of domestic applications of agricultural UAV OA technology, features, hotspot and future research directions. The agricultural UAV OA technology of China is still at an early development stage and many barriers still need to be overcome.

Computer vision is becoming increasingly important for industrial unmanned aerial vehicles (UAVs) to do real-time object and obstacle recognition while they are engaged in autonomous navigation. Variable texture features of objects, on the other hand, frequently lead to feature disappearance, which in turn reduces the accuracy of detection. This work proposed a novel Texture-variant Obstacle Object Classification Model (TOOCM) for feature extraction and dynamic texture representation. By capturing small texture fluctuations in real-time situations, the primary goal is to enhance the accuracy of both obstacle detection and object classification. The TOOCM model incorporates an N-layer ResNet architecture designed to address the difficulties of disappearing features in an adaptable Manner through dynamic layer augmentation based on variations in texture and size. Reconstruction of each convolutional layer in the ResNet is performed on an as-needed basis, taking into consideration the amount of texture concentration in incoming picture regions. The emergence of identifiable texture regions will result in the identification of new objects, which will then trigger adaptive categorization and layer restructuring activities. TOOCM, in contrast to traditional fixed-layer deep models, offers layer-wise learning updates during the navigation of UAV, which guarantees constant performance in complicated settings. The results of the experiments show that TOOCM achieves a greater detection accuracy of 14.09%, enhanced precision of 14.53%, and a loss reduction of 14.17%, particularly in high-density obstacle settings. These results demonstrate that the suggested adaptive feature learning approach is effective.

Dynamic collision avoidance between multiple vessels is a task full of challenges for unmanned surface vehicle (USV) movement, which has high requirements on real-time performance and safety. The difficulty of multi-obstacle collision avoidance is that it is hard to formulate the optimal obstacle avoidance strategy when encountering more than one obstacle threat at the same time; a good strategy to avoid one obstacle sometimes leads to threats from other obstacles. This paper presents a dynamic collision avoidance algorithm for USVs based on rolling obstacle classification and fuzzy rules. Firstly, potential collision probabilities between a USV and obstacles are calculated based on the time to the closest point of approach (TCPA). All obstacles are given different priorities based on potential collision probability, and the most urgent and secondary urgent ones will then be dynamically determined. Based on the velocity obstacle algorithm, four possible actions are defined to determine the basic domain in the collision avoidance strategy. After that, the Safety of Avoidance Strategy and Feasibility of Strategy Adjustment are calculated to determine the additional domain based on fuzzy rules. Fuzzy rules are used here to comprehensively consider the situation composed of multiple motion obstacles and the USV. Within the limited range of the basic domain and the additional domain, the optimal collision avoidance parameters of the USV can be calculated by the particle swarm optimization (PSO) algorithm. The PSO algorithm utilizes both the characteristic of pursuance for the population optimal and the characteristic of exploration for the individual optimal to avoid falling into the local optimal solution. Finally, numerical simulations are performed to certify the validity of the proposed method in complex traffic scenarios. The results illustrated that the proposed method could provide efficient collision avoidance actions.

While it is well known that rain may influence the performance of automotive LIDAR sensors commonly used in ADAS applications, there is a lack of quantitative analysis of this effect. In particular, there is very little published work on physically-based simulation of the influence of rain on terrestrial LIDAR performance. Additionally, there have been few quantitative studies on how rain-rate influences ADAS performance. In this work, we develop a mathematical model for the performance degradation of LIDAR as a function of rain-rate and incorporate this model into a simulation of an obstacle-detection system to show how it can be used to quantitatively predict the influence of rain on ADAS that use LIDAR.

Autonomous navigation in agricultural environments requires precise obstacle classification to ensure collision-free movement. This study proposes a convolutional neural network (CNN)-based model designed to enhance obstacle classification for agricultural robots, particularly in orchards. Building upon a previously developed YOLOv8n-based real-time detection system, the model incorporates Ghost Modules and Squeeze-and-Excitation (SE) blocks to enhance feature extraction while maintaining computational efficiency. Obstacles are categorized as “Real”—those that physically impact navigation, such as tree trunks and persons—and “Fake”—those that do not, such as tall weeds and tree branches—allowing for precise navigation decisions. The model was trained on separate orchard and campus datasets and fine-tuned using Hyperband optimization and evaluated on an external test set to assess generalization to unseen obstacles. The model’s robustness was tested under varied lighting conditions, including low-light scenarios, to ensure real-world applicability. Computational efficiency was analyzed based on inference speed, memory consumption, and hardware requirements. Comparative analysis against state-of-the-art classification models (VGG16, ResNet50, MobileNetV3, DenseNet121, EfficientNetB0, and InceptionV3) confirmed the proposed model’s superior precision (p), recall (r), and F1-score, particularly in complex orchard scenarios. The model maintained strong generalization across diverse environmental conditions, including varying illumination and previously unseen obstacles. Furthermore, computational analysis revealed that the orchard-combined model achieved the highest inference speed at 2.31 FPS while maintaining a strong balance between accuracy and efficiency. When deployed in real-time, the model achieved 95.0% classification accuracy in orchards and 92.0% in campus environments. The real-time system demonstrated a false positive rate of 8.0% in the campus environment and 2.0% in the orchard, with a consistent false negative rate of 8.0% across both environments. These results validate the model’s effectiveness for real-time obstacle differentiation in agricultural settings. Its strong generalization, robustness to unseen obstacles, and computational efficiency make it well-suited for deployment in precision agriculture. Future work will focus on enhancing inference speed, improving performance under occlusion, and expanding dataset diversity to further strengthen real-world applicability.

Rapid boost in the density of the pedestrians and vehicles on the roads have made the life of visually impaired people very difficult. In this direction, we present the design of a smart phone based cost-effective system to guide visually impaired people to walk safely on the roads by detecting obstacles in real-time scenarios. Monocular vision based method is used to capture the video and then frames are extracted out of it after removing the blurriness caused by the motion of camera. For each frame, a computationally simple approach based on the ground plane is proposed for detecting and removing the ground plane. After removing ground plane, features like Speeded-Up Robust Features (SURF) of the non-ground area are computed and compared with features of obstacles. An active contour model is used to segment the area of non-ground image whose SURF features are matched with obstacle features. This area is referred as Region of Interest (ROI). To check whether ROI belongs to an obstacle or not, Gray Level Co-occurrence matrix (GLCM) features are calculated and passed onto a classification model. Classification results show that this system is efficiently able to detect the obstacles that are known to the system in near real-time.

Intelligent driving research has focused much attention on point cloud obstacles since they are a class of high-dimensional data that can adequately depict the shape and placement of obstacles, unlike picture data. Currently, deep learning technology is primarily employed for vehicle autonomy point cloud obstacle classification tasks. These techniques typically struggle with low classification accuracy, processing efficiency, and model stability. To tackle the abovementioned issues, this paper suggests a novel random forest algorithm that integrates the out-of-bag error theory and can consistently and accurately evaluate the influence of point cloud properties. Then, building on the novel algorithm, this paper suggests a modified PointNet network that incorporates the effects of both global and local features on the classification task, therefore increasing the conventional network’s classification accuracy. To assess the effectiveness of this novel approach in the experimental portion, we set up an evaluation system based on the metrics for average accuracy, overall accuracy, and a confusion matrix. According to the simulation results, the overall accuracy of the proposed network in terms of classification accuracy is 94.4% and the average accuracy is 84.9%, which are then compared to the prototype PointNet and its variants. The classification accuracies for the four types of obstacles are 97.6%, 63.6%, 92.5%, and 86.1%. In addition, the proposed method is effective at improving both the computational complexity and stability of the network.

This work introduces a self-supervised architecture for robust classification of moving obstacles in urban environments. Our approach presents a hierarchical scheme that relies on the stability of a subset of features given by a sensor to perform an initial robust classification based on unsupervised techniques. The obtained results are used as labels to train a set of supervised classifiers. The outcomes obtained with the second sensor can be used for higher level tasks such as segmentation or to refine the within-clusters discrimination. The proposed architecture is evaluated for a particular realization based on range and visual information which produces track-based labeling that is then employed to train supervised modules that perform instantaneous classification. Experiments show that the system is able to achieve 95% classification accuracy and to maintain the performance through on-line retraining when working conditions change.

Machine learning techniques have accelerated the development of autonomous navigation algorithms in recent years, especially algorithms for on-road autonomous navigation. However, off-road navigation in unstructured environments continues to challenge autonomous ground vehicles. Many off-road navigation systems rely on LIDAR to sense and classify the environment, but LIDAR sensors often fail to distinguish navigable vegetation from non-navigable solid obstacles. While other areas of autonomy have benefited from the use of simulation, there has not been a real-time LIDAR simulator that accounted for LIDAR–vegetation interaction. In this work, we outline the development of a real-time, physics-based LIDAR simulator for densely vegetated environments that can be used in the development of LIDAR processing algorithms for off-road autonomous navigation. We present a multi-step qualitative validation of the simulator, which includes the development of an improved statistical model for the range distribution of LIDAR returns in grass. As a demonstration of the simulator’s capability, we show an example of the simulator being used to evaluate autonomous navigation through vegetation. The results demonstrate the potential for using the simulation in the development and testing of algorithms for autonomous off-road navigation.

Inspired by the human 3D visual perception system, we present an obstacle detection and classification method based on the use of Time-of-Flight (ToF) cameras for robotic navigation in unstructured environments. The ToF camera provides 3D sensing by capturing an image along with per-pixel 3D space information. Based on this valuable feature and human knowledge of navigation, the proposed method first removes irrelevant regions which do not affect robot’s movement from the scene. In the second step, regions of interest are detected and clustered as possible obstacles using both 3D information and intensity image obtained by the ToF camera. Consequently, a multiple relevance vector machine (RVM) classifier is designed to classify obstacles into four possible classes based on the terrain traversability and geometrical features of the obstacles. Finally, experimental results in various unstructured environments are presented to verify the robustness and performance of the proposed approach. We have found that, compared with the existing obstacle recognition methods, the new approach is more accurate and efficient.