China
Africa
Explore chapters and articles related to this topic
Flight Planning
Published in Yasmina Bestaoui Sebbane, Multi-UAV Planning and Task Allocation, 2020
Aircraft collision is a serious concern as the number of aircraft in operation increases. In the future, they will be expected to carry sophisticated avoidance systems when flying together with conventional aircraft. On-board sensor systems combined with self-operating algorithms will ensure collision avoidance with little intervention from ground stations. On-board sensors can detect other aircraft nearby. Information related to the other aircraft such as position, velocity and heading angle can be used to build an avoidance command. In order for an aircraft to maneuver successfully in such a dynamic environment, a feasible and collision-free trajectory needs to be planned in the physical configuration space. The avoidance law should be generated in real-time and simple to implement. The ability to sense and avoid natural and man-made obstacles and to rebuild its flight path is an important feature that a smart autonomous aircraft must possess [391]. Guidance, trajectory generation, flight and mission planning are the core of the flight management system of a smart autonomous aircraft [52,70]. The computational abilities provide the strongest constraints on the autonomous aircraft, although advances in the hardware mechanisms are to be expected. Improvements in software are essential.
MASS Design and Engineering
Published in R. Glenn Wright, Unmanned and Autonomous Ships, 2020
Autonomous trials of SeaZip 3, an offshore supply ship and crew transfer vessel (SeaZip NL), were held in March 2019 on the North Sea near Den Helder, Netherlands [Russell 2019]. The vessel was outfitted with sensors and software to facilitate several different scenarios for collision avoidance. The vessel was built in 2015 and measures at 168 gross tons, 20 tons deadweight (25.75m length × 10.1m breadth with 2.2m draft, 84ft × 33ft × 7.3ft) with a maximum observed speed of 21.3 knots and average of 20.1 knots [MarineTraffic 2019f]. This is a joint industry project supported by a number of shipping companies, suppliers and universities and is partly funded by the TKI-Maritiem allowance of the Dutch Ministry of Economic Affairs and Climate Policy.
Collision-avoidance under COLREGS for unmanned surface vehicles via deep reinforcement learning
Published in Maritime Policy & Management, 2020
Yong Ma, Yujiao Zhao, Yulong Wang, Langxiong Gan, Yuanzhou Zheng
Corresponding to the above traditional collision-avoidance techniques, with the development of artificial intelligence, especially the emerged machine learning methods, recently several studies have tried to resolve collision-avoidance issues by employing deep learning (DL) algorithm and reinforcement learning (RL) algorithm. By using deep reinforcement learning (DRL) system, Wang (Wang et al. 2019) constructed a collision-avoidance decision system. Wherein the optimal motion state was obtained based on Markov decision process, and the effects of collision-avoidance was evaluated through a designed reward function. Cheng (Cheng and Zhang, 2018) employed DRL to resolve anti-collision problem for USVs with static obstacles, and numerical simulation verified that DRL can resolve multi-obstacle avoidance issue under simple scenarios.