Journal of Field Robotics最新文献

Path planning is crucial for autonomous robot navigation and operation. Tasks like cleaning, inspection, and mining, all require complete coverage operation. For maps of convex regions, a reciprocating coverage method can be used. However, for maps of concave shapes, it is unsuitable. For this purpose, this paper proposes an image-based map segmentation method for complete coverage path planning. Taking the grip map as an image, it is used to divide a concave map into convex subregions. For each convex region, it will generate a batch of waypoints for the robot controller. The subregions are then connected to achieve a complete coverage of the entire region. On the basis of a global path planning, a local path following, and real-time obstacle avoidance methods, the complete coverage operation is achieved. Moreover, a coverage ratio calculation method is proposed and shown real-timely in a visual interface. Extensive experiments in simulations and real-world environments demonstrate the effectiveness of this method, achieving an average coverage ratio of 97.89% and a maximum of 92.19% in the presence of obstacles. Most importantly, this method has been successfully tested on an autonomous mining vehicle, achieving an average coverage ratio of 96% in given maps.

LiDAR-based SLAM is recognized as one effective method to offer localization guidance in rough environments. However, off-the-shelf LiDAR-based SLAM methods suffer from significant pose estimation drifts, particularly components relevant to the vertical direction, when passing to uneven terrains. This deficiency typically leads to a conspicuously distorted global map. In this article, a LiDAR-based SLAM method is presented to improve the accuracy of pose estimations for ground vehicles in rough terrains, which is termed Rotation-Optimized LiDAR-Only (ROLO) SLAM. The method exploits a forward location prediction to coarsely eliminate the location difference of consecutive scans, thereby enabling separate and accurate determination of the location and orientation at the front-end. Furthermore, we adopt a parallel-capable spatial voxelization for correspondence-matching. We develop a spherical alignment-guided rotation registration within each voxel to estimate the rotation of vehicle. By incorporating geometric alignment, we introduce the motion constraint into the optimization formulation to enhance the rapid and effective estimation of LiDAR's translation. Subsequently, we extract several keyframes to construct the submap and exploit an alignment from the current scan to the submap for precise pose estimation. Meanwhile, a global-scale factor graph is established to aid in the reduction of cumulative errors. In various scenes, diverse experiments have been conducted to evaluate our method. The results demonstrate that ROLO-SLAM excels in pose estimation of ground vehicles and outperforms existing state-of-the-art LiDAR SLAM frameworks.

Magnetic miniature soft robots hold significant potential in biomedical research, especially for targeted therapy, drug delivery, and cell manipulation. Precise path tracking control is crucial for these robots in complex biomedical applications. Here, we propose a Stanley path tracking control algorithm based on visual feedback for magnetic soft robots. First, a magnetic miniature soft crawling robot was designed and fabricated, and its crawling mechanism was detailed. Next, a simulation framework using the material point method (MPM) was constructed to simulate the movement and deformation of the miniature robot and to verify the proposed crawling mechanism. Finally, visual feedback technology was used to obtain the robot's position and posture, and the Stanley algorithm was applied for path tracking control in crawling mode. The effectiveness of the proposed path tracking control strategy has been verified through multiple experiments. Compared with the traditional Pure Pursuit control method, it has higher robustness and better control accuracy.

Corn row-controlled weeding is a critical crop field management aspect. Corn row-controlled weeding robots suffer from large errors in seedling and weed identification algorithms, lack of row weeding actuators in dry fields, and low integration of automated devices for identification, navigation, and weeding. Therefore, this paper investigates an intelligent robot for row-controlled weeding, which could realize the integrated automatic operation of seedling and weed identification, row line acquisition, automatic navigation, and row-controlled weeding. We present a fully integrated, autonomous, and innovative solution for row-controlled weeding robots to overcome the difficulties of weeding to row. The solution constructs a seedling and weed identification model based on the YOLOv5 model and an improved boundary loss function. It also investigates a real-time extraction method for corn seedling strips based on region-of-interest updates. In addition, we developed an intelligent control device that can realize row-controlled weeding and depth control, adjusting weed height, depth of entry, and weed spacing. Finally, a fully autonomous weeding robot system was developed and integrated. Field tests showed that the intelligent robot could continuously cruise autonomously for weeding, with a weeding rate higher than 79.8% and a seedling injury rate lower than 7.3%. These efforts have laid a solid foundation for the future commercialization of intelligent weeding robots.

Reforestation is pivotal in mitigating climate change, preserving biodiversity, and safeguarding ecosystems. Precision reforestation implies efficiently using materials and human resources to conduct this intensive task. In this sense, using unmanned aerial vehicles (UAVs), also known as drones, improves the accuracy of dispensed seeds while increasing coverage and reducing labor. However, drone-based reforestation still presents technological challenges that need to be addressed. An important challenge is the presence of disturbances during flights due to environmental conditions, primarily unexpected wind, and the unavoidable loss of mass presented by the vehicle caused by the dispensing task. This paper evaluates the use of an observer-based controller to reject the particular disturbances caused by the sowing activity. Through meticulous experimentation and analysis, the study demonstrates the observer's adeptness in mitigating external disturbances, thereby enhancing the precision and stability of UAV operations. This technological advancement holds promise for diverse practical applications and has implications for environmental conservation efforts, particularly reforestation. The obtained experimental results confirm the viability of the proposed controller and observer framework, highlighting its potential to improve the robustness of environmental monitoring, conservation, and sustainable resource management practices.

The high efficiency of multiple quadrotors coordination makes it widely used. However, the technology faces threats from internal collisions as well as external obstacles. This paper proposes a multiquadrotors distributed cooperative collision-free tracking control framework, which consists of three parts: collision avoidance mechanism, obstacle avoidance mechanism and tracking control. First, in the face of internal and external threats, the collision avoidance function with damping is designed based on Hooke's law. And the potential energy function is designed for multiple quadrotors obstacle avoidance based on bird flock obstacle avoidance. Then, a distributed control protocol is proposed based on consensus theory. Finally, a multi-quadrotors simulation and experiment platform with the same architecture is built. We deploy the framework on the multiple quadrotor platform and conduct collision avoidance, obstacle avoidance, and encirclement experiments with 30 quadrotors. The experimental results show that the quadrotors can perform the task well. The minimum distance between quadrotors is 0.8022 m, and the minimum distance from obstacles is 0.8866 m, all of which meet the safety distance requirements. Moreover, compared with classical and advanced methods in simulation, our proposed method has the smallest average tracking error, only 0.3019 m, and improves task time by 7.03% and 6.23%, respectively, which verifies the effectiveness and practicality of the proposed framework of distributed cooperative collision avoidance and tracking control for multiple quadrotors.

The deployment of the robotic system that executes specific task is being challenged by the prevalence of dynamic objects in real-world scenes. Two robotic tasks sparked by this challenge, known as dynamic Simultaneous Localization and Mapping (SLAM) and moving object perception, are crucial for enhancing system robustness and reinforcing environment awareness. Existing public datasets are diverse in platforms, sensor combinations, scenarios, and label annotations, but few adequately benchmark the above tasks. To fill this gap, we introduce the multimodal and campus-scapes (M2CS) dataset, providing robot-centric synchronized LiDAR-Inertial-Visual-GNSS data with 3D moving object annotation in specific dynamic scenarios. The dataset exhibits variation in dynamic object types and densities, annotating over 160,000 Light Detection and Ranging (LiDAR) scans and releasing ground truth of trajectories acquired by the GNSS-RTK/INS system. The dataset evaluates existing SLAM and moving object perception methods, driving relevant research to overcome this challenge. We publish the M2CS dataset on the website (https://github.com/Zhaohuanfeng/M2CS) and hope it promotes research on robotics in complex environment.

Ancient architecture embodies the culmination of historical building techniques and artistic expression, representing a valuable heritage of history, art, and technology. These buildings not only document the cultural traditions and architectural evolution of a nation but also preserve significant aspects of human civilization. However, over time, ancient buildings gradually deteriorate due to both natural and human factors. The issue of cracks is particularly critical in the preservation of ancient buildings. Cracks not only affect the esthetic appeal of these structures, but also, if left unaddressed, they can lead to irreversible damage. Existing technologies struggle to address both the marking of defect locations and the calculation of defect information. For example, image recognition technology can identify cracks in a photo, but it is unable to determine the specific location of the crack within the building, nor can it calculate the three-dimensional information of the crack. To address this, we combined point cloud technology with crack detection algorithms to develop a novel method. First, we integrated point cloud data acquired from terrestrial laser scanning (TLS) and supplementary remotely piloted aircraft (RPA) data to construct a comprehensive point cloud model of the building for archiving. Next, we conduct point cloud density analysis on the model to extract crack regions based on density variations and then analyze these regions to determine crack locations and compute detailed information. To validate this method, we conducted experiments on a 600-year-old wooden building on our campus as a case study. The experimental results indicate that this method can accurately determine the specific location of cracks, with the calculated three-dimensional information corresponding to their actual positions. This method has also proven to be reliable for continuous annual monitoring, allowing for the ongoing detection and analysis of changes in cracks over time.