Velocity and Acceleration Estimation and Uniform Hardness Perception in 6-DOF Haptic Rendering

With advances in human computer interface technologies, the sense of touch has been successfully brought into many virtual reality applications. Haptic devices are such kinds of interfaces which can generate programmable forces to its users, allowing for natural manipulations of objects inside virtual environments. In this thesis, we study two critical issues with haptic interfaces that have a tool-like end effector.;The second issue is about the influences of a haptic interface's effective mass and viscous damping on users' perceived virtual stiffness in 6-DOF haptic rendering. In any haptic simulations, we always hope that the desired force can be delivered to the user faithfully. But the inherent dynamics of the device may have negative effects on the final perceived force. Through experiments, we have found that the induced forces from those physical parameters affect the perceived virtual stiffness non-uniformly: a surface that should feel uniformly stiff will feel softer or harder, depending on which part of the virtual tool it contacts. The scale of the non-uniform perception is also related to the user's holding point as well as its relative position to the contact points on the tool. Nevertheless, the goal of this study is not to increase the perceived hardness but to make users perceive a uniform hardness when tapping a virtual object. For that, we propose three different methods to compensate for the non-uniform hardness perception. The subjective evaluations show the effectiveness of all three methods.;The first issue is about velocity and acceleration estimation for haptic interfaces. Since most haptic devices are equipped with neither tachometers nor accelerometers, this information has to be derived from discrete position measurements. However, due to a limited sensor resolution and a fast update rate requirement of haptic simulations, getting accurate and robust estimates is challenging. In studies of human arm/hand trajectories, researchers observed that skilled movements are usually extremely smooth that is equivalent to minimizing an objective measure with a performance index of jerk. With the assumption of a constrained minimum-jerk model for human arm/hand trajectories, an adaptive constrained 4-state Kalman filter is proposed for the velocity and acceleration estimation for haptic devices.