IEEE Trans Haptics. 2012;5(3):264-73. doi: 10.1109/TOH.2012.36.
Haptic interaction between a human leader and a robot follower in waltz is studied in this paper. An inverted pendulum model is used to approximate the human's body dynamics. With the feedbacks from the force sensor and laser range finders, the robot is able to estimate the human leader's state by using an extended Kalman filter (EKF). To reduce interaction force, two robot controllers, namely, admittance with virtual force controller, and inverted pendulum controller, are proposed and evaluated in experiments. The former controller failed the experiment; reasons for the failure are explained. At the same time, the use of the latter controller is validated by experiment results.
本文研究了人类领导者和机器人跟随者在华尔兹中的触觉交互。使用倒立摆模型来近似人类的身体动力学。通过力传感器和激光测距仪的反馈，机器人可以使用扩展卡尔曼滤波器（EKF）来估计人类领导者的状态。为了减少相互作用力，提出并评估了两种机器人控制器，即导纳与虚拟力控制器和倒立摆控制器。前者控制器的实验失败了；解释了失败的原因。同时，通过实验结果验证了后者控制器的使用。
