Intelligent Mission Planning and Tracking Control of Autonomous Surface Vehicles Based on Neural Computation

Description

Autonomous surface vehicles (ASVs) are marine vessels capable of performing various marine operations with minimal or without human supervision. Due to their salient features of high autonomy and mobility, they are vitally important tools for numerous applications in marine environment monitoring, hydrographic survey, ocean exploration and exploitation, search and rescue missions, etc.Compared with their counterparts for land and aerial vehicles, the mission planning and motion control of ASVs are more challenging due to the nonlinearity, uncertainty and underactuation of marine vehicle dynamics as well as the strong disturbances of waves and tides between liquid and air media at sea. For complex missions, there are increasing needs for deploying a fleet of ASVs instead of a single one to complete difficult tasks. Cooperative operations among multiple ASVs offer great advantages with enhanced capability and efficiency. Despite various application potentials, mission planning and motion control of swarmed ASVs pose great challenges due to the multiplicity of ASVs, complexity of intra-vehicle interactions and fleet formation with collision avoidance requirements, and scarcity of communication bandage at sea environments. High-performance mission planning and motion control methods are deemed to be highly demanded for widespread applications of ASVs.In recent years, neural computation regained its popularity from academia and industries, due to the numerous successes of deep learning. In addition, the results of our recent research findings in an RGC-funded project indicate that collaborative neurodynamic approaches are effective for distributed and global optimization. These outcomes may play vital roles in coordinated planning and control of ASVs. Systematic investigations on neural-computation-based intelligent mission planning and motion control would be highly interesting and rewarding.In this proposed research, an in-depth investigation of neural-computation-based intelligent ASV mission planning and motion control methodologies will be performed. The research will consist of four coherent parts. The first part will focus on developing neurodynamic approaches to centralized and distributed ASV task assignment in dynamic and uncertain operational environments. The second part will focus on developing collaborative neurodynamic optimization methods for receding-horizon trajectory generation of ASV fleets. The third part will aim at developing intelligent motion control methods based on neural computation. The last part will aim at experimentation and demonstration of the intelligent mission planning and motion control methods on an experimental platform. It is expected that the successful completion of the proposed project will significantly advance the research and development of ASV mission planning and motion control methodologies for both academia and industry.