基于模型预测控制的NAO机器人长廊地形稳定行走

Corridor Navigation for NAO Humanoid Robot Based on Model Predictive Control

室内长廊移动导航是移动机器人领域的经典问题,然而由于关节运动偏差和脚底打滑等因素,NAO仿人机器人内置的标准直线行走功能却无法直接实现长廊行走。本文以实现NAO机器人在长廊环境中的稳定前向行走为具体目标,提出了一种包含视觉伺服、运动规划,步态控制的集成式行走导航框架和具体方法。本文首先采用基于图像的视觉伺服控制算法生成期望运动速度,以此确定步行方向和足迹序列跟随输入的运动速度,并利用基于模型控制(MPC)的在线步态生成算法处理双足步行的ZMP约束和稳定性约束,进而生成符合步行运动特征并满足预设约束的质心轨迹。本文通过实验实现了NAO机器人靠近长廊中心地稳定双足行走,既验证了所提方法的有效性,也为足式机器人的高性能室内行走提供了基础方法与参照。

Navigation in Indoor corridor is a classic problem in mobile robot research,due to factors like the joint movement deviation and feet slippage,the standard straight-line motion module built in NAO cannot fulfill corri-dor walking.This paper aims to achieve stable forward walking of NAO in the corridor environment,we presented an integrated navigation system consisting with visual servoing,motion planning and gait control that allows humanoid robot NAO to autonomously navigate in the corridor environments.Firstly,the image-based controller for visual servoing generates the motion command,then walking direction and footprint determined to track the desired input command,at the core of gait control architecture is a Model Predictive Control(MPC)for dealing with ZMP and stability constraints of biped walking,this method can generates a stable center-of-mass trajectory that accord with characteristics of biped walking and satisfy various constraints.Experimental results for the humanoid robot NAO walking in the corridor are presented to show the validity of the method,and also provides references for high-performance indoor walking of legged robots.