In the future, it may be possible to employ large numbers of autonomous marine vehicles to perform tedious and dangerous tasks, such as minesweeping. Hypothetically, groups of vehicles may leverage their numbers by co...
详细信息
ISBN:
(纸本)0780385438
In the future, it may be possible to employ large numbers of autonomous marine vehicles to perform tedious and dangerous tasks, such as minesweeping. Hypothetically, groups of vehicles may leverage their numbers by cooperating. A fundamental form of cooperation is to perform tasks while maintaining a geometric formation. The formation behavior can then enable other cooperative behaviors. In this paper, we describe a leader-follower formation-flying control algorithm. This algorithm can be applied to one-, two-, and three-dimensional formations, and contains a degree of built-in robustness. Simulations and experiments are described that characterize the performance of the formation control algorithm. The experiments utilized surface craft that were equipped with an acoustic navigation and communication system, representative of the technologies that constrain the operation of underwater autonomous vehicles. The simulations likewise included the discrete-time nature of the communication and navigation.
This study applies path planning and obstacle avoidance to drone control for conducting riverbank inspections. Given that the river's surrounding environments are often windy and filled with overgrown branches and...
详细信息
This study applies path planning and obstacle avoidance to drone control for conducting riverbank inspections. Given that the river's surrounding environments are often windy and filled with overgrown branches and unknown obstacles, this study improves path planning and obstacle avoidance to enable drones to complete inspection tasks using the planned optimal route. Multiple drones are used for larger-scale areas to reduce time consumption and increase efficiency. Regarding path planning, the A* algorithm is improved using a grid-based approach. For obstacle avoidance, depth cameras are installed on the drones, and the obtained images are processed by reinforcement Q-learning with a genetic algorithm to navigate around obstacles. Since maintaining formation is necessary during task execution, the leader-follower method of formation flight ensures that multiple drones can complete inspection tasks while maintaining formation.
Various control algorithms have been developed for fleets of autonomous vehicles. Many of the successful control algorithms in practice are behavior-based control or nonlinear control algorithms, which makes analyzing...
详细信息
ISBN:
(纸本)9780791842164
Various control algorithms have been developed for fleets of autonomous vehicles. Many of the successful control algorithms in practice are behavior-based control or nonlinear control algorithms, which makes analyzing their stability difficult. At the same time, many system theoretic approaches for controlling a fleet of vehicles have also been developed. These approaches usually use very simple vehicle models such as particles or point-mass systems and have only one coordinate system which allows stability to be proven. Since most of the practical vehicle models are six-degree-of-freedom systems defined relative to a body-fixed coordinate system, it is difficult to apply these algorithms in practice. In this paper, we consider a formation regulation problem as opposed to a formation control problem. In a formation control problem, convergence of a formation from random positions and orientations is considered, and it may need a scheme to integrate multiple moving coordinates. On the contrary, in a formation regulation problem, it is not necessary since small perturbations from the nominal condition, in which the vehicles are in formation, are considered. A common origin is also not necessary if the relative distance to neighbors or a leader is used for regulation. Under these circumstances, the system theoretic control algorithms are applicable to a formation regulation problem where the vehicle models have six degrees of freedom. We will use a realistic six-degree-of-freedom model and investigate stability of a fleet using results from decentralized control theory. We will show that the leader-follower control algorithm does not have any unstable fixed modes if the followers are able to measure distance to the leader. We also show that the leader-follower control algorithm has fixed modes at the origin, indicating that the formation is marginally stable, when the relative distance measurements are not available. Multi-vehicle simulations are performed using a hybr
暂无评论