Optimal Control for Multiple Unmanned Underwater Crawling Vehicles

摘要

This paper presents the development of an autonomous group behavior using Unmanned Underwater Crawling Vehicles (UUCV) to perform a mission of searching for targets in an unstructured underwater mission field. The mission is completed through the task of detection and gathering information of target-like objects, and the task of target identification. The behavior-based control in which a behavior is characterized by a sensory information is adapted for the creation of the formation of UUCVs to maximize the efficiency of the detection task of target like objects in the mission field. This group behavior is highly group leader-centered formation control, but this does not imply a centralized formation control in which the commands of all the followers are generated by the designated group leader. There are three different repertoire of behaviors used in the behavior based control of multiple UUCVs. One behavior is the selection mechanism of a group leader. The others are the Leader-follower formation using an intelligent fuzzy logic agent to determine the command of a follower and group leader under the given formation constraints in a decentralized manner, and an obstacle avoidance algorithm with the sonar and contact sensors built in a bumper system. The Optimal Multiple Paths Generation Algorithm (OMPGA) is applied to perform the process of task allocation in heterogeneous unmanned group behavior as well as to generate multiple paths used in the task of target identification in the optimal process. OMPGA which is conducted by a leader vehicle in the group, is a kind of constraint multi-objective optimal algorithm using a Neural Network that embeds the evolutionary algorithm. The group behavior of UUCVs to perform the target search mission is demonstrated in a simulation environment. The simulation environment is modeled by using the is designed as well as the contact dynamics between UUCV and surface of mission.