High-speed online trajectory planning for UAVs poses a significant challenge due to the need for precise modeling of complex dynamics while also being constrained by computational limitations. This paper presents a multi-fidelity reinforcement learning method (MFRL) that aims to effectively create a realistic dynamics model and simultaneously train a planning policy that can be readily deployed in real-time applications. The proposed method involves the co-training of a planning policy and a reward estimator; the latter predicts the performance of the policy's output and is trained efficiently through multi-fidelity Bayesian optimization. This optimization approach models the correlation between different fidelity levels, thereby constructing a high-fidelity model based on a low-fidelity foundation, which enables the accurate development of the reward model with limited high-fidelity experiments. The framework is further extended to include real-world flight experiments in reinforcement learning training, allowing the reward model to precisely reflect real-world constraints and broadening the policy's applicability to real-world scenarios. We present rigorous evaluations by training and testing the planning policy in both simulated and real-world environments. The resulting trained policy not only generates faster and more reliable trajectories compared to the baseline snap minimization method, but it also achieves trajectory updates in 2 ms on average, while the baseline method takes several minutes.
Zhejiang University(浙江大学);Huzhou Institute, Zhejiang University(浙江大学湖州研究院);National Engineering Research Center for Industrial Automation (Ningbo Institute)(工业自动化国家工程研究中心(宁波研究院))
Perception-aware Planning for Quadrotor Flight in Unknown and Feature-limited Environments
School of Intelligence Science and Engineering, Harbin Institute of Technology(哈尔滨工业大学智能科学与工程学院);Department of Mechanical and Energy Engineering, Southern University of Science and Technology(南方科技大学机械与能源工程系)
Shape-Adaptive Planning and Control for a Deformable Quadrotor
State Key Laboratory of Industrial Control Technology, Zhejiang University(浙江大学工业控制技术状态重点实验室);Huzhou Institute, Zhejiang University(浙江大学湖州市院);Faculty of Robot Science and Engineering, Northeastern University(东北大学机器人科学与工程学院)
Motion-Specific Battery Health Assessment for Quadrotors Using High-Fidelity Battery Models
Department of Convergence IT Engineering, Pohang University of Science and Technology(convergence IT工程系,釜山科学技术大学);Department of Electrical Engineering, Pohang University of Science and Technology(电气工程系,釜山科学技术大学)
APACE: Agile and Perception-Aware Trajectory Generation for Quadrotor Flights
School of Artificial Intelligence, Sun Yat-Sen University(中山大学人工智能学院);Department of Electronic and Computer Engineering, Hong Kong University of Science and Technology(香港科技大学电子与计算机工程学系)
EGO-Swarm: A Fully Autonomous and Decentralized Quadrotor Swarm System in Cluttered Environments
Zhejiang University(浙江大学);National Engineering Research Center for Industrial Automation (Ningbo Institute)(国家工业自动化工程技术研究中心(宁波研究院));Huzhou Institute, Zhejiang University(浙江大学湖州研究院)
Minimum-Time Quadrotor Waypoint Flight in Cluttered Environments
University of Zurich(苏黎世大学);ETH Zurich(苏黎世联邦理工学院)
我们解决的问题,规划一个最小的时间轨迹的四旋翼在一系列指定的航路点在障碍物的存在,同时利用完整的四旋翼动力学。这个问题对于自主搜索和救援以及无人机竞赛场景是至关重要的,但是到目前为止,机器人社区\emph{in its entirety}还没有解决这个问题,这是因为在避免碰撞所造成的非凸约束的存在下,最小化时间的挑战。早期的工作依赖于简化的动力学或多项式轨迹表示,没有利用四旋翼的全部致动器潜力,因此,没有以最小化时间为目标。我们解决这个具有挑战性的问题,通过使用一个分层的,基于采样的方法与一个越来越复杂的四旋翼模型。我们的方法首先发现路径在不同的拓扑结构,以指导后续的轨迹搜索的运动动力学点质量模型。然后,它使用一个渐进最佳的,基于Kinodynamic采样的方法,基于一个完整的四旋翼模型上的点质量的解决方案,找到一个可行的轨迹与时间最优的目标。所提出的方法被证明在杂乱的环境中优于所有相关的基线,并在世界上最大的运动捕捉系统之一的实际飞行中以超过60公里/小时的速度进一步验证。我们将代码开源。
GapFlyt: Active Vision Based Minimalist Structure-less Gap Detection For Quadrotor Flight
University of Maryland Institute for Advanced Computer Studies(马里兰大学高级计算机研究所)
虽然四旋翼和一般的空中机器人本质上是主动的,但到目前为止,它们在文献中的感知能力大多是被动的。今天的研究人员和从业者使用传统的计算机视觉算法,目的是建立一个普遍适用的表示:场景的3D重建。使用这种表示,规划任务的构建和完成,让四旋翼演示自主行为。这些方法是低效的,因为它们不是任务驱动的,并且这些方法不被飞行的昆虫和鸟类利用。这些智能体已经解决了导航和复杂控制的问题,而不需要构建3D地图,并且高度任务驱动。在本文中,我们提出了这个框架的四旋翼生物启发的感知设计。我们使用这种理念来设计一个极简的感觉电机框架,使四旋翼飞行器能够在不使用单目相机和机载传感的情况下通过未知的间隙飞行。我们在许多具有不同设置和窗口形状的真实实验中成功评估和演示了所提出的方法,即使最小公差仅为5 cm,在2.5 ms $^{-1}$下也实现了85%的成功率。据我们所知,这是第一篇论文,它解决了一个未知的形状和位置的单目相机和板载传感的间隙检测的问题。
Time-Optimal Planning for Long-Range Quadrotor Flights: An Automatic Optimal Synthesis Approach
University of Toronto Institute for Aerospace Studies(多伦多大学航空航天研究所);University of Toronto Robotics Institute(多伦多大学机器人研究所);University of Toronto Division of Engineering Science(多伦多大学工程科学系);Technical University of Munich(慕尼黑工业大学);Munich Institute of Robotics and Machine Intelligence (MIRMI)(慕尼黑机器人与机器智能研究所)
Rotor-Failure-Aware Quadrotors Flight in Unknown Environments
School of Robotics and Automation, Nanjing University(南京大学机器人与自动化学院);Institute of Cyber-Systems and Control, College of Control Science and Engineering, Zhejiang University(浙江大学控制系统研究所);Department of Aeronautical and Aviation Engineering, The Hong Kong Polytechnic University(香港理工大学航空工程系);School of Aeronautic Science and Engineering, Beihang University(北航航空科学与工程学院)
GS-Planner: A Gaussian-Splatting-based Planning Framework for Active High-Fidelity Reconstruction
Institute of Cyber-Systems and Control(cyber系统与控制研究所);College of Control Science and Engineering(控制科学与工程学院);Zhejiang University(浙江大学);Huzhou Institute, Zhejiang University(浙江大学湖州研究院)
Aggressive Trajectory Generation for A Swarm of Autonomous Racing Drones
Zhejiang University(浙江大学);Zhejiang University of Technology(浙江工业大学);Northwestern Polytechnical University(西北工业大学)
自主无人机竞赛正在成为挑战四旋翼自主技术的绝佳平台,包括规划,导航和控制技术。然而,关于这个主题的大多数研究主要集中在单个无人机场景。在本文中,我们描述了一种新的时间最优轨迹生成方法,用于生成时间最优轨迹,使一群四旋翼飞行器以其最大机动性无碰撞地通过预定义的航路点。我们在Gazebo仿真中验证了该方法,其中5个四旋翼的群可以在35 m * 35 m的空间中以14 m/s的最高速度飞行通过复杂的6路点赛道。在4 m × 2 m的飞行平台上,对两个四旋翼飞行器进行了通过3个航路点的飞行试验,验证了该方法的可行性。仿真和实际飞行测试表明,该方法可以为一群自主竞赛无人机生成最优攻击轨迹。该方法也可以很容易地转移到其他类型的机器人群。
FLYINGTRUST: A Benchmark for Quadrotor Navigation Across Scenarios and Vehicles
School of Electronics and Communication Enginnering, Sun Yet-sen University, Shenzhen, China(电子与通信工程学院,孙逸森大学,深圳,中国);Tianwei Xunda (Hunan) Technology Co., Ltd., Hunan, China(天威湘电(湖南)科技有限公司,湖南,中国)