里程计
底盘
运动学
卡尔曼滤波器
控制理论(社会学)
扩展卡尔曼滤波器
计算机科学
线性化
协方差
噪音(视频)
非线性系统
控制工程
人工智能
工程类
机器人
数学
移动机器人
航空航天工程
控制(管理)
物理
图像(数学)
统计
经典力学
量子力学
作者
Qingfeng Lou,Francisco González,József Kövecses
标识
DOI:10.1109/lra.2019.2895393
摘要
State estimation is crucial for exploration rovers. It provides the pose and velocity of the rover by processing measurements from onboard sensors. Classical wheel odometry only employs encoder measurements of the two wheels in the differential drive. As a consequence, input noise can lead to large uncertainties in the estimated results. Also, the estimation models used in classical wheel odometry are nonlinear, and the linearization process that propagates the mean and covariance of the estimated state introduces additional errors in the process. This letter puts forward a novel wheel odometry approach for six-wheeled rovers. A kinematic model is formulated to relate the velocity of the wheels and the chassis, and later used to develop the corresponding estimation model. The components of the velocity of the chassis, decomposed in the chassis-fixed coordinate frame, are selected as the system state in the estimation, which results in a linear model. The motions of all wheels are fused together to provide the measurements. Wheel slip is considered random Gaussian noise in this kinematic model. The continuous-time Kalman filter is employed to process the model. Experimental validation with six-wheeled rover prototypes was used to confirm the validity of the proposed approach.
科研通智能强力驱动
Strongly Powered by AbleSci AI