Stability and Performance Analysis of Centralized and Distributed Multi-rate Control Architectures for Multi-user Haptic Interaction

This paper is concerned with multi-user haptic simulation environments in which users can interact across an Ethernet-based Local Area Network (LAN) or a Metropolitan Area Network (MAN). Using network protocols such as the UDP and TCP/IP under normal network traffic conditions, the achievable real-time packet communication rate can be well below the 1 kHz update rate suggested in the literature for high fidelity haptic rendering. However by adopting a multi-rate control strategy, the local control loops can be executed at a much higher rate than that of the data packet transmission between the user workstations. Within such a framework, two control architectures, namely centralized and distributed are presented. Mathematical models of the controllers are developed and used in a comparative analysis of their stability and performance. The results of such analysis demonstrate that the distributed control architecture has greater stability margins and outperforms the centralized controller. It is also shown that the limited network packet transmission rate can degrade the haptic fidelity by introducing a viscous damping into the perceived impedance of the virtual object. Using the proposed models, this damping value is calculated and compensated by active control. Experiments conducted with a dual-user/dual-finger haptic platform confirm the analytical results.