基于类脑计算的深空探测器自主技术理论和方法研究

自主运行是深空探测器关键技术之一，如何将类脑智能运用到深空探测自主技术中，研究出更高效自主的、更低资源消耗的自主运行控制技术是人工智能与航天技术交叉研究前沿。本课题构建基于大规模GPU集群的类脑计算仿真平台，针对深空探测环境的不确定性和高复杂性，开展基于类脑计算的深空探测器自主技术研究，包括环境感知与认知技术、自主学习多任务控制技术、故障诊断与自修复技术等，进而开展上述类脑计算模型算法的性能评估与优化方法研究。本课题研究探索类脑计算模型下更高效、安全、可靠深空探测器自主技术，为实现层次化、模块化深空探测器自主运行管理系统提供理论方法。

Autonomous operation is one of the key technologies of deep space probe. How to apply brain-like intelligence to the autonomous technology of deep space probe, and how to develop more efficient autonomous and lower resource consumption autonomous operation control technology is the research frontier of the intersection of artificial intelligence and space technology. In this project, brain-like computing simulation platform will be constructed based on a large-scale GPU cluster, for the uncertainty and high complexity of deep space exploration environment, and the research on autonomous technology of deep space detector based on brain-like computing will be carried out, including environment perception and cognition technology, autonomous learning multi task control technology, fault diagnosis and self-repair technology, etc. and then the performance of the above brain-like computing model algorithm will be carried out research on evaluation and optimization methods. In this paper, we explore the more efficient, safe and reliable autonomous technology of deep space probe under the brain-like computing model, and provide a theoretical method for the realization of hierarchical and modular autonomous operation management system of deep space probe.