Safety Assured Online Guidance With Airborne Separation for Urban Air Mobility Operations in Uncertain Environments
Safety Assured Online Guidance With Airborne Separation for Urban Air Mobility Operations in Uncertain Environments
复制标题
DOI：
10.1109/tits.2022.3163657
复制
发表时间：
2022-10
影响因子：
8.5
通讯作者：
Pengcheng Wu;Xuxi Yang;Peng Wei;Jun Chen
中科院分区：
文献类型：
--
作者：
Pengcheng Wu;Xuxi Yang;Peng Wei;Jun Chen
城市空中交通（UAM）的概念提出了使用革命性的新型电动垂直起降（eVTOL）飞机在以前航空市场服务不足的地方之间提供高效和按需的航空运输服务。如何在高密度、动态和不确定的空域环境中管理大规模自主飞行操作并保证安全是无人机管理成功的关键。针对不确定性条件下的多机自主飞行问题，提出并分析了一种具有机载自分离能力的安全分散在线制导算法。该问题被制定为一个多智能体马尔可夫决策过程与连续的行动空间，并解决了一个定制的分散在线算法的基础上蒙特卡洛树搜索（MCTS）。为保证不确定环境下实时自主飞行操作的安全性，引入机会损失约束分离公式，并将其与MCTS算法相结合。此外，高斯过程回归沿着与贝叶斯优化离散化的连续动作空间，这有助于缩短飞行时间。仿真结果表明，该算法能够在不确定的高密度空域环境中提供安全的机载制导，并保证较低的近空中碰撞概率。
The concept of Urban Air Mobility (UAM) proposes to use revolutionary new electrical vertical takeoff and landing (eVTOL) aircraft to provide efficient and on-demand air transportation service between places previously underserved by the current aviation market. A key challenge for the success of UAM is how to manage large-scale autonomous flight operations with safety guarantee in high-density, dynamic and uncertain airspace environments. In this paper, a safety assured decentralized online guidance algorithm with airborne self-separation capability is proposed and analyzed for multi-aircraft autonomous flight operations under uncertainties. The problem is formulated as a multi-agent Markov Decision Process with continuous action space and is solved by a customized decentralized online algorithm based on Monte Carlo Tree Search (MCTS). To guarantee the safety of real-time autonomous flight operations in uncertain environments, the formulation of loss of chance constrained separation is introduced and integrated with the proposed MCTS algorithm. In addition, Gaussian process regression along with Bayesian optimization is employed to discretize the continuous action space, which helps shorten the flight time. A comprehensive numerical study shows that the proposed algorithm can provide safe onboard guidance with guaranteed low near mid-air collision probability in uncertain and high-density airspace environments.