Force sensor modeling and friction compensation for haptic interfaces

Haptic interfaces are used to provide touch feedback to a user from a virtual (or remote) environment. Impedance control strategies involve measuring motions from the haptic interface and commanding forces to it. Accurate force rendering is degraded by the mechanical impedance of the haptic interface mechanism. A model of this impedance is required to properly compensate for its effects. This thesis presents advancements in haptic interface modeling and in compensation techniques.; The modeling advancements relate to the calculation of an interaction force vector. The haptic interface at The University of Colorado (CU) measures individual rod forces in each degree of freedom. Past analyses have combined the force measurements from these rods into a 5-vector using geometric considerations only. This thesis shows that this approach is only accurate in the limit where the lateral impedance of these rods is insignificant compared to the axial impedance. A new model of the haptic interface is presented herein. This new model is of the same form as an existing model, yet the output from this new model has a different physical interpretation; the output is a sensed force 5-vector not an interaction force. This new model is then augmented to show how an interaction force could be calculated. An estimate is provided of how different these two force vectors typically are for the haptic interface at CU. The discovery of the previous model inaccuracy has implications for haptic interface design and control.; The new compensation technique combines existing methodolgies of motion-based control and force feedback. Combining these two methods can provide substantial friction mitigation with a computationally simple controller. Unlike some previous compensation techniques for a 5 degree-of-freedom haptic interface, the new hybrid technique does not require the calculation of a model inverse. The hybrid technique has been implemented on the CU haptic interface using a second-order diagonal force controller and a simple smoothed Coulombic friction model. While this controller doesn't perform well according to the typical control metrics such as bandwidth, it is able to significantly reduce a user's perception of the mechanism friction when emulating free space conditions.