Article
Robotics
Zhefan Xu, Di Deng, Kenji Shimada
Summary: The proposed dynamic exploration planner (DEP) utilizes incremental sampling and Probabilistic Roadmap (PRM) to explore unknown environments, demonstrating successful exploration in dynamic environments while outperforming benchmark planners in terms of exploration time, path length, and computational time.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2021)
Article
Robotics
Ping Zhong, Bolei Chen, Siyi Lu, Xiaoxi Meng, Yixiong Liang
Summary: This paper proposes an information-driven exploration strategy for Unmanned Aerial Vehicles (UAVs) in unknown environments, utilizing the fast marching method. The strategy includes frontier point detection, evaluation of candidate goals based on a utility function, and optimization of UAV trajectory and yaw angle. Simulation experiments demonstrate the superiority of the proposed strategy.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Computer Science, Artificial Intelligence
Hub Ali, Gang Xiong, Muhammad Husnain Haider, Tariku Sinshaw Tamir, Xisong Dong, Zhen Shen
Summary: This paper presents a trajectory planning technique for global and local path planning of a UAV in 3D terrain. It proposes a feature selection-based decision model and an A* multi-directional planner with an extensive search area to generate optimal global paths. Offset trajectories are also generated for local path planning to avoid collisions. The simulation results show that this approach outperforms other 3D UAV path planning techniques.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Environmental Sciences
Jiajun Ou, Xiao Guo, Wenjie Lou, Ming Zhu
Summary: The study proposes a deep reinforcement learning-based framework for safe autonomous navigation of quadrotors in semi-known environments. The framework utilizes dueling double deep recurrent Q-learning for global path planning and contrastive learning-based feature extraction for real-time autonomous obstacle avoidance. Experimental results show remarkable performance in both global path planning and autonomous obstacle avoidance.
Article
Remote Sensing
Tzu-Jui Lin, Karl A. Stol
Summary: Modern plantation forest procedures often rely on manual data acquisition, which limits the quantity and quality of the collected data. This study presents an autonomous system using multi-rotor UAVs to explore plantation forest environments. The proposed method, including waypoint selection, trajectory generation, and trajectory following, is tested extensively in simulation and real flight testing, demonstrating its robust performance.
Article
Robotics
Jiajie Yu, Hao Shen, Jianyu Xu, Tong Zhang
Summary: This letter proposes an efficient heuristic viewpoint determination method for autonomous exploration, which addresses the low efficiency issue. By randomly generating higher-quality initial viewpoints using a Gaussian sampler, selecting the next viewpoint with a fresh heuristic evaluation function, and refining the viewpoint, the proposed method outperforms the state-of-the-art frontier-based method by 15%-25% in almost all scenarios, as indicated by extensive simulations and real-world tests.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Review
Remote Sensing
Amber Israr, Zain Anwar Ali, Eman H. Alkhammash, Jari Juhani Jussila
Summary: This survey investigates the latest research on motion planning for UAVs using bio-inspired algorithms, and summarizes the contributions and limitations of each article.
Article
Engineering, Civil
Jinchao Chen, Chenglie Du, Ying Zhang, Pengcheng Han, Wei Wei
Summary: Unmanned aerial vehicles (UAVs) are widely utilized in civilian and military applications for their high autonomy and strong adaptability. This paper addresses the coverage path planning problem of autonomous heterogeneous UAVs on a bounded number of regions by proposing an exact formulation based on mixed integer linear programming and a clustering-based algorithm inspired from density-based clustering methods to achieve optimal flight paths and efficient coverage tasks. Experiments demonstrating the efficiency and effectiveness of the proposed approach with randomly generated regions are conducted.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Automation & Control Systems
Xinyu Cai, Shane Kyi Hla Win, Hitesh Bhardwaj, Shaohui Foong
Summary: In this study, a novel modular aerial robotic platform called ARROWs is introduced, which can be easily reconfigured with customized wing and control modules. Unlike conventional multirotor aerial vehicles, ARROWs generate more lift through revolving wings. However, the complex dynamics pose challenges in flight controller development. To address this, a cascaded flight controller is designed based on simplified flight dynamics and relaxed hovering conditions, while inertial measurement units are employed to estimate flight configuration. Experimental results validate the proposed platform and flight control strategy in 12 different configurations.
IEEE-ASME TRANSACTIONS ON MECHATRONICS
(2023)
Article
Robotics
Ruibin Zhang, Yuze Wu, Lixian Zhang, Chao Xu, Fei Gao
Summary: This paper presents an autonomous and adaptive navigation framework for terrestrial-aerial bimodal vehicles, enabling complete autonomy and improved performance through the generation of safe and low-power trajectories in unknown environments and dynamic energy consumption adjustments. Real-world experiments demonstrate the framework's robustness and energy-saving capabilities.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Chemistry, Analytical
Youkyung Hong, Sunggoo Jung, Suseong Kim, Jihun Cha
Summary: This study proposes a complete hardware and software architecture for autonomous area coverage missions using multiple UAVs, optimizing waypoint allocation to achieve efficient and autonomous mission execution.
Article
Environmental Sciences
Yongwoo Lee, Junkang An, Inwhee Joe
Summary: This paper proposes a method to create a new flight route during the autonomous flight of an unmanned aerial vehicle by self-recognition and judgment. By effectively utilizing the hardware resources of small computers, objects can be quickly and accurately recognized, and quick detection and avoidance can be achieved through filtering and object resizing.
Article
Computer Science, Information Systems
Vojtech Spurny, Vaclav Pritzl, Viktor Walter, Matej Petrlik, Tomas Baca, Petr Stepan, David Zaitlik, Martin Saska
Summary: This paper presents a novel approach to autonomous extinguishing of indoor fires inside a building by a Micro-scale Unmanned Aerial Vehicle (MAV), discussing key technologies and evaluating the performance of the entire system. The system presented is part of a complex multi-MAV solution and is currently being used as the core of a fire-fighting Unmanned Aerial System (UAS) industrial platform.
Article
Mathematics
Faten Aljalaud, Heba Kurdi, Kamal Youcef-Toumi
Summary: This paper presents a novel path planning heuristic inspired by the booby bird's foraging behavior for multi-UAV pipe inspection missions. The heuristic enables each UAV to find an optimal path for defect detection in pipe networks while avoiding collisions. The proposed method outperforms existing algorithms in terms of mean detection time and computational efficiency under different settings.
Article
Chemistry, Analytical
Zhijian Li, Wendong Zhao, Cuntao Liu
Summary: This paper studies a UAV-UGV-enabled data collection system, proposes two cooperative strategies, and designs a collaborative strategy selection algorithm to optimize data collection time.
Article
Computer Science, Artificial Intelligence
Jonathan Spencer, Sanjiban Choudhury, Matthew Barnes, Matthew Schmittle, Mung Chiang, Peter Ramadge, Sidd Srinivasa
Summary: This study introduces the Expert Intervention Learning (EIL) method, which learns collision avoidance for robots in real and simulated driving tasks through expert interventions. The approach can learn from just a few hundred samples (about one minute) of expert control.
Article
Robotics
Peng Yin, Ivan Cisneros, Shiqi Zhao, Ji Zhang, Howie Choset, Sebastian Scherer
Summary: This article introduces iSimLoc, a learning-based global relocalization approach that can match visual data with significant appearance and viewpoint differences. The method utilizes a place recognition network to match query images to reference images of different stylistic domains and viewpoints. The hierarchical global relocalization module enables fast and accurate pose estimation. iSimLoc achieves high successful retrieval rates and outperforms other methods in terms of inference time.
IEEE TRANSACTIONS ON ROBOTICS
(2023)
Article
Robotics
Shiqi Zhao, Peng Yin, Ge Yi, Sebastian Scherer
Summary: We propose SphereVLAD++, an attention-enhanced viewpoint invariant place recognition method, which projects point clouds onto a spherical perspective and captures contextual connections between local features and global 3D geometry distribution. It outperforms all relative state-of-the-art 3D place recognition methods, achieving successful retrieval rates of 7.06% and 28.15% under small or even totally reversed viewpoint differences. It also has low computation requirements and high time efficiency, making it suitable for low-cost robots.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Robotics
Yafei Hu, Junyi Geng, Chen Wang, John Keller, Sebastian Scherer
Summary: This letter proposes a method to learn the state value function for guiding robot exploration in real-world challenging environments. It combines offline Monte-Carlo training and online Temporal Difference adaptation to optimize the trained value estimator. An intrinsic reward function based on sensor information coverage is also designed to enable the robot to gain more information. The results show that our method achieves better prediction and exploration performance compared with existing approaches.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Engineering, Electrical & Electronic
Zhipeng Zhao, Huai Yu, Chenwei Lyu, Wen Yang, Sebastian Scherer
Summary: Visual localization plays a crucial role in intelligent robots and autonomous driving, especially when GNSS is unreliable. To address the limited information provided by the commonly used pinhole camera with a narrow field-of-view, we propose an end-to-end learnable network that correlates 360 degrees spherical images with point clouds for cross-modal visual localization. Inspired by the attention mechanism, we optimize the network to capture salient features for comparing images and point clouds. Our approach is evaluated on the KITTI-360 dataset and achieves promising results.
IEEE SENSORS JOURNAL
(2023)
Article
Robotics
Peng Yin, Shiqi Zhao, Haowen Lai, Ruohai Ge, Ji Zhang, Howie Choset, Sebastian Scherer
Summary: In this article, a novel framework called AutoMerge is introduced for merging large-scale maps. This framework is capable of accurately associating and optimizing large-scale map segments without using GPS, and it can detect and discard incorrect loop closures caused by perceptual aliasing. By utilizing techniques such as multiperspective fusion and adaptive loop closure detection, AutoMerge achieves accurate data associations and performs pose graph optimization to smooth the merged map.
IEEE TRANSACTIONS ON ROBOTICS
(2023)
Article
Robotics
Jingtian Yan, Xingqiao Lin, Zhongqiang Ren, Shiqi Zhao, Jieqiong Yu, Chao Cao, Peng Yin, Ji Zhang, Sebastian Scherer
Summary: This paper presents a new approach for multi-agent exploration in a bounded 3D environment with unknown initial poses. The proposed method intelligently directs one agent to repeat another agent's trajectory based on the quality indicator of sub-map merging. Experimental results demonstrate that this approach improves exploration efficiency by up to 50% compared to baselines while maintaining robust sub-map merging.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Proceedings Paper
Automation & Control Systems
Brady Moon, Satrajit Chatterjee, Sebastian Scherer
Summary: This article introduces a sampling-based approach to address the problem of informative path planning in high-dimensional spaces and non-trivial sensor constraints. The method utilizes informed sampling and considers potential information gain to generate an optimal path and identify objects of interest in large search spaces. It outperforms the baseline method by 18.0% in experiments.
2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS)
(2022)
Proceedings Paper
Automation & Control Systems
Mohammadreza Mousaei, Junyi Geng, Azarakhsh Keipour, Dongwei Bai, Sebastian Scherer
Summary: This paper presents a study on the fault tolerance of tiltrotor VTOL aircraft regarding various actuator failures. By designing and modeling a customized tiltrotor VTOL UAV, analyzing its feasible wrench space, and designing dynamic control allocation, the system can adapt to actuator failures and maintain controlled flight.
2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS)
(2022)
Proceedings Paper
Automation & Control Systems
Seungchan Kim, Chen Wang, Bowen Li, Sebastian Scherer
Summary: This research proposes a human-interactive framework for detecting human-informed interesting objects through few-shot online learning. An unsupervised learning algorithm is applied on the unmanned vehicle to recognize interesting scenes, and annotations provided by a human operator are used to learn and detect similar objects.
2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS)
(2022)
Proceedings Paper
Automation & Control Systems
Yao He, Huai Yu, Wen Yang, Sebastian Scherer
Summary: This paper presents a Visual-Inertial Odometry (VIO) algorithm with multiple non-overlapping monocular cameras to improve the robustness of the VIO algorithm. The algorithm utilizes multiple cameras to capture more stable features, enabling stable state estimation even when some cameras track unstable or limited features. To address the high CPU usage caused by multiple cameras, a GPU-accelerated frontend is proposed. Experimental results using a pedestrian carried system demonstrate that the multi-camera setup significantly improves estimation robustness without increasing CPU usage.
2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Bowen Li, Chen Wang, Pranay Reddy, Seungchan Kim, Sebastian Scherer
Summary: Few-shot object detection has gained increasing attention and progress, but the requirement of offline fine-tuning stage hinders its usage in online applications. The proposed AirDet architecture achieves comparable or better results without fine-tuning by learning class-agnostic relation with support images.
COMPUTER VISION, ECCV 2022, PT XXXIX
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Chen Wang, Yuheng Qiu, Dasong Gao, Sebastian Scherer
Summary: This paper proposes a method that combines graph neural networks (GNN) with lifelong learning by converting a continuous graph learning problem to a regular graph learning problem. The authors introduce a new topology called feature graph, which treats features as new nodes and converts nodes into independent graphs. This method shows efficiency and superior performance in various applications.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Proceedings Paper
Computer Science, Artificial Intelligence
Nikhil Varma Keetha, Chen Wang, Yuheng Qiu, Kuan Xu, Sebastian Scherer
Summary: Object encoding and identification are crucial for robotic tasks, and previous approaches have limitations in representing objects from multiple viewpoints and dealing with unknown objects. This paper proposes a novel temporal 3D object encoding method called AirObject, which generates global object embeddings using structural information and graph attention mechanism. It demonstrates superior performance in video object identification compared to existing methods.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Article
Engineering, Electrical & Electronic
Taimeng Fu, Huai Yu, Wen Yang, Yaoyu Hu, Sebastian Scherer
Summary: In this article, a targetless cross-modal calibration system is presented for extrinsic calibration among stereo cameras, thermal cameras, and laser sensors. The calibration is done by minimizing registration error and optimizing edge feature alignment. The method does not rely on dedicated targets and allows for multisensor calibration in a single shot.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)