陆上无人平台自主决策与规划技术综述
A Review of Decision-making and Motion Planning Technology for Unmanned Ground Platforms
付梦印 1宋文杰 2张婷 2毛梓豪 2李德伦2
作者信息
- 1. 北京理工大学自动化学院,北京100081;南京理工大学自动化学院,南京210094
- 2. 北京理工大学自动化学院,北京100081
- 折叠
摘要
陆上无人平台自主导航技术是其实现智能化、协同化运行的核心,而自主决策与规划技术是自主导航系统在复杂动态交互、集群协同冲突等挑战下智能、安全、高效、平稳工作的关键.以单无人平台智能决策规划和集群平台协同决策规划为切入点,从技术演变和应用场景等角度回顾并论述了相关文献.在单平台决策规划方面,围绕基于规则方法到数据驱动模型的演变进程,重点讨论了动态交互场景下的时空联合规划以及大模型自主学习的特性.并以此延伸到复杂任务与环境约束下的集群决策规划策略,分别从集中式、分布式、机器学习三个方向对现有文献进行综合论述.最后,系统总结了现有技术的发展现状,并对该领域未来发展方向进行了展望.
Abstract
Autonomous navigation is the core technology to achieve intelligence and collaborative operation for un-manned ground platforms, while decision-making and planning are crucial for the systems to work intelligently, safely, effi-ciently, and smoothly in complex and interactive scenarios. Firstly, starting from the intelligent decision-making planning of single unmanned platform and collaborative decision-making planning of cluster platform, relevant literatures are reviewed and discussed from the perspectives of technological evolution and application scenarios. For single-platform decision-mak-ing planning, the evolution from rule-based methods to data-driven planning models is delved, the spatiotemporal joint planning and large-scale language model autonomous learning characteristics in dynamic interactive scenarios is mainly dis-cussed. Next, the scope to collaborative planning in complex tasks and environment constraints is extended, an overview of existing methods, from the directions of centralized, distributed, and deep learning-based approaches are provided. Final-ly, the current technology status and potential future directions are summarized.
关键词
自动驾驶/行为决策/运动规划/大模型/集群协同规划Key words
autonomous driving/decision-making/motion planning/large-scale language model/cluster collaborative planning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据