Journal of Field Robotics最新文献

The cover image is based on the article Tracking mooring lines of floating structures by an autonomous underwater vehicle by Sehwa Chun et al., 10.1002/rob.70076.

The image illustrates a jet-powered personal aerial system (PAV) performing vertical take-off and landing within a post-disaster urban environment. The system integrates five micro turbojet engines and a dual-degree-of-freedom vector control mechanism to achieve high maneuverability, stability, and fault tolerance. This work demonstrates the feasibility of a compact, human-scale VTOL platform capable of safe operation in complex field conditions, with potential applications in rescue and rapid response missions.

This study presents a novel method for tracking mooring lines of Floating Offshore Wind Turbines (FOWTs) using an Autonomous Underwater Vehicle (AUV) equipped with a tilt-controlled Multibeam Imaging Sonar (MBS). The proposed approach enables the AUV to estimate the 3D positions of mooring lines and safely track them in real-time, overcoming the limitations of traditional Remotely Operated Vehicle (ROV)-based inspections. By utilizing the tilt-controlled MBS and a pre-trained You Only Look Once (YOLO) model, the AUV identifies the mooring lines within sonar imagery and dynamically adjusts its velocities to maintain a safe distance during the inspection. A re-navigation method using the Rauch-Tung-Striebel (RTS) smoother enhances the AUV's localization accuracy by correcting its trajectory using post-processed data from sensors such as the Doppler Velocity Log (DVL), Super Short Baseline (SSBL) system, and Global Navigation Satellite System (GNSS). Additionally, reconstruction with catenary curve fitting is employed to estimate the mooring line's catenary parameters, offering insights into its potential deformation. The approach was validated using a hovering-type AUV, Tri-TON through both tank experiments and a sea experiment at an FOWT Hibiki in Kitakyushu, Japan. In the sea experiment, the AUV successfully tracked the mooring lines for 423 s, demonstrating its ability to estimate the position and catenary parameters of the mooring lines. The experimental results highlight areas for future improvement, particularly in enhancing localization accuracy, developing robust control algorithms, and expanding the analysis of mooring line conditions. This method lays the groundwork for future advancements in automated mooring line inspections and enables the integration of additional techniques, such as visual inspection.

Time-correlated single-photon counting (TCSPC) lidar enables high-resolution 3D imaging at kilometer range. While previous works have covered long range TCSPC 3D imaging from either stationary or airborne platforms, this is, to the best of our knowledge, the first attempt that use a moving ground vehicle. Fusing measurements taken at different locations and time imposes very high demands on knowledge of the sensor pose which are hard to achieve on such platforms. In this study we use simultaneous localization and mapping (SLAM) to correct for positioning errors, allowing us to create high fidelity point clouds using long range TCSPC lidar imaging on a moving ground vehicle. Our method uses inertial and GNSS sensors to get an initial estimate of the sensor motion, which is used to reconstruct parts of the scene over short time intervals. The initial motion estimate is then refined by adding constraints from local point cloud matching, and the refined estimate is used to construct the final point cloud of the target area. We describe the sensor system and integration of all sensors, as well as the field trial at which the system was evaluated. The proposed method is able to generate a high fidelity point cloud of a wooded target area from a distance of roughly 800 m while measuring from a moving vehicle. Compared to measurements from a stationary position we obtain better coverage of the target area, and increased ability to penetrate into the forest. However, some precision is lost in the reconstructed point cloud.

We present a comprehensive teleoperation framework for electric vehicle (EV) battery cell handling, integrating haptic feedback, extended reality (XR) visualization, and task-parameterized Gaussian mixture regression (TP-GMR) for adaptive, real-time trajectory generation. The system enables seamless switching between manual and autonomous operation through a variable autonomy mechanism, while constraint barrier functions (CBFs) enforce spatial safety constraints. A lightweight intent prediction module anticipates user deviation and precomputes corrective trajectories, reducing response time from 2.0 s to under 1 ms. The framework is implemented on an industrial KUKA robotic manipulator and validated in structured and real-world EV battery disassembly scenarios. Results show that combining XR and haptic feedback reduces task completion time by up to 48% and path deviation by 32%, compared to manual teleoperation without assistance. Predictive replanning improves continuity of force feedback and reduces unnecessary user motion. The integration of XR-based spatial computing, learning-from-demonstration, and real-time control enables safe, precise, and efficient manipulation in high-risk environments. This study demonstrates a scalable human-in-the-loop solution for battery recycling and other semi-structured tasks, where full automation is impractical. The proposed system significantly improves operator performance while maintaining safety and flexibility, marking a meaningful advancement in collaborative field robotics.

Typical landing gears of small uninhabited aerial vehicles (UAV) limit their capability to land on vehicles moving at more than 20–50 km/h due to high drag forces, high pitch angles and potentially high relative horizontal velocities. To enable landing at higher speeds, a combination of lightweight friction shock absorbers and reverse thrust was developed. This allows for rapid descents (i.e., 3 m/s) toward the vehicle while leveling at the last instant. Simulations show that the proposed system is (1) more robust at higher descent speeds contrary to traditional configurations, (2) can touchdown at almost any time during the leveling maneuver, thus reducing the timing constraints, and (3) is robust to many environmental, design and operational factors, maintaining a success rate above 80% up to 100 km/h. Compared to standard multirotors, this approach expands the possible state envelope at touchdown by a factor of 60. A total of 38 experimental trials were conducted where a drone successfully landed on a pickup truck moving at speeds ranging from 10 to 110 km/h. The increased touchdown envelope was shown to improve the multirotors' robustness to external disturbances such as winds and wind gusts, sensor errors and unpredictable motion of the ground vehicle. The increased landing capabilities also expand the flight envelope at the start of the leveling maneuver by a factor of 38 compared to a standard multirotor, thereby allowing the drone to fly in tougher conditions and initiate its leveling maneuver from a broader range of altitudes, vertical and horizontal velocities, as well as pitch angles and rates.