无人水面车辆发展与挑战：综合回顾

需积分: 47 88 浏览量更新于2024-07-18 7 收藏 1.66MB PDF 举报

"这篇英文文章是对无人船技术的综合回顾，涵盖了无人船的历史发展、先进技术、导航与控制系统，以及未来面临的挑战。" 在当前科技快速发展的时代，无人船（Unmanned Surface Vehicles, USVs）作为海洋探索和作业的重要工具，正受到商业、科学和军事领域的广泛关注。随着对海洋及浅水区域研究的深入，对于具备高级制导、导航和控制（Guidance, Navigation, and Control, GNC）能力的USVs的需求日益增强。这篇文章对近年来USVs的发展进行了详尽的文献综述。文章首先概述了USVs的发展历程，从早期的基础设计到现代的高科技集成，展示了USVs如何逐渐演变成能够自主执行任务的先进系统。作者提到了一些基本概念和定义，帮助读者理解无人船技术的基础框架。接着，文章深入探讨了现有的USVs GNC方法，这些方法包括但不限于路径规划、避障策略、定位技术、自主控制算法等。通过对各种应用、方法和挑战的分类，作者揭示了这些技术在实际应用中的复杂性和多样性。GNC系统对于USVs的精确操作和自主性至关重要，它确保了无人船能在不同环境和任务中有效工作。此外，文章还强调了未来USVs面临的一般性挑战。这些挑战可能涉及技术层面，如提高传感器精度、增强通信能力以实现远程控制和数据传输，以及提升能源效率以延长作业时间。同时，也涉及法规和伦理问题，因为无人船的广泛应用需要与国际海洋法和安全规定相协调。在自主性方面，文章指出，实现更高程度的自主决策和学习能力将是USVs未来发展的一个关键方向。这包括利用人工智能和机器学习技术，使USVs能够适应不断变化的环境，自我修复故障，并做出更加智能的决策。这篇年度评论提供了无人船技术的全面视图，不仅回顾了过去的发展，还对未来可能的方向进行了展望。对于研究者、工程师和对该领域感兴趣的读者来说，这是一个深入了解USVs现状和未来趋势的重要资源。通过持续的技术创新和克服挑战，无人船有望在海洋科学研究、环境保护、海上安全和军事任务等多个领域发挥更大作用。

Z. Liu et al. / Annual Reviews in Control 41 (2016) 71–93 75

3.1. Path planning

As the fundamental aspect of USV guidance systems, path plan-

ning can generally be distinguished between the global and local

approaches. From the literature, a broad spectrum of eﬃcient and

intelligent path planning techniques for USVs are identiﬁed in the

following.

3.1.1. Global path planning

1. Optimization methods : As an attractive method, advanced opti-

mization method (OM) can directly produce optimal trajectories

or paths that might include sophisticated characteristics, such

as spatiotemporal-optimal, danger level (collision probability),

fuel saving, weather routing, formation control, and scheduled

missions.

Inspired by the behavior of biological systems, evolutionary al-

gorithm (EA) represent a class of artiﬁcial intelligence increas-

ingly employed in the design of USV path planners. EA can be

characterized into an optimization problem with speciﬁed con-

straints. Genetic algorithm (GA) are to date the most widely

adopted method for waypoints generation ( Campbell et al.,

2012 ). In Svec and Gupta (2011, 2012) , a strongly-typed genetic

programming (GP)-based evolutionary approach is developed to

enable USVs to protect a target from intruder boats.

Due to their expensive computational costs, especially when

constraints such as obstacles, USV dynamic limits, and mission

constraints must be satisﬁed, optimization methods are limited

for real-time implementation.

2. Heuristic search algorithms : Application of heuristic approaches

in path planning ﬁrst appeared in the early 20 0 0s. The A

∗

search algorithm is a widely used grid-based strategy, which

can quickly ﬁnd an optimal path with the least number of

nodes. But the drawbacks of large computational memory costs

and unwanted sharp turns result in diﬃculty in its practical ap-

plication in the cases where quick and real-time control of USVs

are necessary.

In Larson, Bruch, and Ebken (2006) , the A

∗

search algorithm

is chosen to ﬁnd an optimal path within a limited amount of

time. Another application is presented in Naus and Wa

(2013) ,

where the A

∗

algorithm is employed to search the shortest and

safest trajectory in an electronic navigational chart. A

∗

and lo-

cally bounded optimal planning under uncertainty are com-

bined in Svec, Schwartz, Thakur, and Gupta (2011) . In Zhuang,

Su, Liao, and Sun (2012) , a marine radar image-based local path

planning method is developed for USVs. The path is searched

by Dijkstra’s algorithm which is a special case of the A

∗

algo-

rithm. To improve performance with respect to paths and com-

putational consumption, a modiﬁed version of the A

∗

algorithm

named as direction priority sequential selection (DPSS) is ap-

plied in Naeem, Irwin, and Yang (2012a) . In Svec, Thakur, Shah,

and Gupta (2012) , a combination of a model-predictive and an

∗

based algorithm is introduced. An A

∗

based curvature path

planning algorithm is proposed in Kim, Park, and Myung (2013) ,

where both the actual turning capability of USVs, and the en-

vironmental information, such as the terrain, buoy and fairway,

are explicitly considered in the cost map.

3.1.2. Local path planning

1. Line-of-sight : A successful guidance technique that is widely

employed in missile guidance, line-of-sight (LOS) methods are

equally valid for USVs ( Annamalai & Motwani, 2013; Breivik

et al., 2008; Caccia, Bibuli, Bono, & Bruzzone, 2008a; Cac-

cia et al., 2005; Desa, Maurya, Pereira, Pascoal, & Prabhude-

sai, 2007; Fredriksen & Pettersen, 2006; Naeem, Sutton, & Xu,

2012b; Peng, Wang, Chen, Hu, & Lan, 2013; Sharma & Sutton,

2013; Tran, Choi, Baek, & Shin, 2014; Xu, Sutton, & Sharma,

2007 ). There are also modiﬁed versions for better real-time im-

plementation such as biased-LOS ( Naeem et al., 2012a ). Despite

these, there are still drawbacks associated with LOS, including:

(1) the potential to overshoot, caused by the environmental dis-

turbances compensating action ( Campbell et al., 2012 ), and (2)

the connection between the waypoints still being rigid lines,

even though paths have been smoothed to some extent.

2. Potential ﬁelds : As deﬁned in the potential ﬁeld (PF) approach,

the objectives are assigned with attractive ﬁelds, while obsta-

cles are distributed with repulsive ﬁelds. USVs are thus moving

toward the attractive ﬁelds and away from the repulsive ﬁelds

( Khatib, 1986 ). Although this method is characterized by its ef-

fective implementation and low computational consumption in

real-time, it is normally only employed for local path planning

since it is prone to guide USVs to local minima instead of the

objectives ( Campbell et al., 2012 ).

An interesting implementation of a PF-based USV path plan-

ning scheme is introduced in Healey, Horner, Kragelund, Wring,

and Monarrez (2007) . In Soltan, Ashraﬁuon, and Muske (2009) ,

obstacles are approximately enclosed by ellipse ﬁelds which

are the solutions of a class of ordinary differential equations

(ODEs). However, much effort is still required to overcome the

local-minima problem by employing depth-ﬁrst and best-ﬁrst

search techniques, wavefront-based strategies, or harmonic po-

tential functions.

3.1.3. Hybrid path planning

In order to ensure the safe and effective path planning for USV

moving among the preliminarily speciﬁed or dynamically chang-

ing waypoints in the dynamic and hazardous environment, increas-

ing efforts have recently been dedicated to the hybrid path plan-

ning strategy that consists of both global and local path plan-

ning approaches. In Larson et al. (2006) and Larson, Bruch, Hal-

terman, Rogers, and Webster (2007) , a hybrid path planning ap-

proach is presented, which is constituted by a two-layered hier-

archical architecture combining with both global and local path

planning functions. The A

∗

algorithm is employed to ﬁnd a feasi-

ble solution for the global path planning, while a behavior-based

common world model approach is utilized to manage the near-

ﬁeld changes that arise to the previously deﬁned path, so that the

USV can keep following the preplanned path. Casalino, Turetta, and

Simetti (2009) presents a multi-layered hierarchical architecture, a

global path is computed in the ﬁrst layer based on the Dijkstra al-

gorithm, while the second layer modiﬁes this predeﬁned path in a

locally optimal way adopting the A

∗

method. In Svec et al. (2012) , a

lattice-based hybrid path planning method is developed with con-

sideration of the USV dynamics.

3.2. Path replanning

As the major role of path replanning, collision avoidance ( Yu

& Zhang, 2015 ) is generally overlooked in the basic guidance laws

(an obstacle-free path is commonly assumed). Unfortunately, re-

cent statistics show that 60% of casualties at sea are caused by

collisions ( Naeem et al., 2012a ). Obstacles, such as lobster traps,

buoys, ﬁshing nets, submerged rocks, other maritime traﬃc, new

constructions, variable water levels, and sea debris, can all po-

tentially contribute to collision risks. Without the ability to avoid

collisions, USVs may collide with any objects present along the

planned path. In addition, a collision avoidance module can also

enhance the autonomy of USVs to avoid approaching objects by

conducting autonomous path replanning.

3.2.1. Protocol-free collision avoidance

In Soltan, Ashraﬁuon, and Muske (2011) , obstacles are as-

sumed to be enclosed by elliptical shapes, and a set of ordinary

剩余22页未读，继续阅读

会飞的小弋

粉丝: 26
资源: 10

无人水面车辆发展与挑战：综合回顾

船只定位.zip

无人船中英文文献.zip

无人船控制系统（部分）

无人驾驶汽车综述论文.pdf

无人车算法综述

无人船.zip_PID STM32_stm32 无人船代码_stm32 算法_stm32定位_无人船代码

无人驾驶车辆轨迹跟踪控制综述型参考英文文献！

无人船控制系统设计

python 无人船

ardupilot无人船

最新资源