A Joint-space Position Control-based Approach to Haptic Rendering of Stiff Objects using Gain Scheduling

Haptic rendering often deals with interactions between stiff objects. A traditional way of force computing models the interaction using a spring-damper system, which suffers from stability issues when the desired stiffness is high. Instead of computing a force, this paper continues to explore shifting the focus to rendering an interaction with no penetration, which can be accomplished by using a position controller in the joint space using the encoders as feedback directly. In order to make this approach easily adaptable to any device, an alternative way to model the dynamics of the device is also presented, which is to linearize a detailed simulation model. As a family of linearized models is used to approximate the full dynamic model of the system, it is important to have a smooth transition between multiple sets of controller gains generated based on these models. Gain scheduling is introduced to improve the performance in certain areas and a comparison among three controllers is conducted in a simulation setup.

Automated summary

This paper explores a new approach to haptic rendering by shifting the focus to rendering interactions with no penetration, using a position controller in the joint space. It also introduces gain scheduling to improve performance and compares three controllers in a simulation setup. Additionally, an alternative way to model the dynamics of the device by linearizing a detailed simulation model is presented to make the approach adaptable to any device.