基于语音交互的四足机器人多模式控制实验设计  

Multimodal Control Experimental Design of Quadruped Robot Based on Natural Language Voice Interaction

作　　者：陈腾 荣学文[1] 李贻斌[1] CHEN Teng;RONG Xuewen;LI Yibin(School of Control Science and Engineering,Shandong University,Jinan 250061,China)

机构地区：[1]山东大学控制科学与工程学院,济南250061

出　　处：《实验室研究与探索》2024年第10期65-69,106,共6页Research and Exploration In Laboratory

基　　金：山东省自然科学基金项目(ZR2022QF027);山东省高等学校青创科技支持计划项目(2023KJ029)。

摘　　要：针对四足机器人全向运动控制、步态与行为切换、人员跟随、自主导航等多模式运动中的人机交互难题,设计了1套基于语音识别与指令解析的四足机器人交互控制系统。明确了四足机器人语音交互系统的硬件组成,并搭建了包含语音采集、识别、解析、播报一体的交互模块;基于SOFM网络设计了语音识别算法;基于ROS建立了语音模块与机器人运动控制模块的通信交互机制,实现了语音指令下达与机器人状态信息播报反馈。通过一系列四足机器人语音交互控制实验,验证了在动态场景下机器人的语音识别准确率高于85%,而在静态场景下则超过95%,保障了四足机器人不同运动模式的准确运动控制。Aiming at the human-computer interaction problems in the multi-mode motion of quadruped robots,such as omnidirectional locomotion control,gait and behavior switching,personnel following,and autonomous navigation,a quadruped robot interactive control system based on speech recognition and command parsing is designed.The hardware components of the quadruped robot’s voice interaction system are clarified,and an interactive module including voice collection,recognition,analysis,and broadcasting is built.A voice recognition algorithm is designed based on the SOFM network.A communication and interaction mechanism between the voice module and the robot locomotion control module is established based on ROS.The system realizes the issuance of voice commands and the broadcasting and feedback of robot status information.Through a series of quadruped robot voice interaction control experiments,it is verified that the robot’s voice recognition accuracy is higher than 85%in dynamic scenes,and higher than 95%in static scenes,ensuring accurate locomotion control of the quadruped robot in different motion modes.

关 键 词：四足机器人 语音交互 神经网络 运动控制 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]