视觉感知和步态控制——国际自主智能机器人大赛U型赛算法综述

Visual perception and gait control---an overview of the algorithm of U-shaped competition in the international competition of autonomous running robots

近年来,随着人工智能技术的发展,可独立完成任务的自主智能机器人成为研究热点。视觉感知和步态控制作为机器人完成各项任务的重要手段,是亟需不断深入研究的技术点。为了拓展小型人形机器人的自主智能水平和应用场景,本文从国际自主智能机器人大赛技术问题出发,研究分析视觉感知和步态控制在模拟人类生活场景的任务(即U型赛)中的应用。首先,对比赛环境和软硬件平台进行介绍,其次分别分析了视觉感知和步态控制在大赛中的应用方法和算法实现,并以上下开横杆为例,介绍了U型赛参赛队伍所使用的算法并进行分析比较。随后,展示了历年参赛队伍的优秀成果,分析了比赛任务完成度,指出比赛的研究难点及可能的解决方向。最后,对未来大赛的技术应用发展方向进行总结。

In recent years,with the development of artificial intelligence technology,autonomous intelligent robots that can independently complete tasks have become a research hotspot.Visual perception and gait control,as important means for robots to complete various tasks,are technology points that need to be continuously deepened and improved.In order to expand the autonomous intelligence of small humanoid robots,this paper explores the applications of visual perception and gait control in tasks that simulate human life scenarios(i.e.,the U-shape competition)from the perspective of the International Competition of Autonomous Running Robots.Specifically,this paper first provides a brief introduction to the competition environment and software and hardware platforms,and then analyzes the applications and commonly used classical algorithms of visual perception and gait control in the competition.Next,this paper takes the up and down pole as an example,introduces the algorithms used by the participating teams in the Ushape competition,and conducts analysis and comparison.This paper also shows the excellent achievements of the participating teams over the years,analyzes the task success of the competition,and points out the research challenges and possible solutions.Finally,this paper summarizes the development direction of future competitions.