Return to search

Modelling and Control of Cooperative Multi-Master/Multi-Slave Teleoperation Systems

<p> Cooperative teleoperation combines two traditional areas of robotics, i.e. teleoperation
and collaborative manipulation. Cooperative telerobotic systems consist of multiple pairs of master I slave robotic manipulators operating in a shared environment. Due to dynamic interaction among slave manipulators as well as communication latency, control of such systems is particularly challenging and the application of standard teleoperation controller may result in instability. </p> <p> In this thesis a multilateral control framework is proposed for cooperative teleoperation systems that allows for transmission of position and force information between all master and slave robots rather than merely between corresponding units. Two different control approaches are introduced that establish kinematic correspondence among masters and slaves. The operators are presented with a virtual intervening tool in order to collaboratively interact with the environment. Models of operators, master and slave robots, tool, and environment are incorporated in the design. </p> <p> A multilateral adaptive nonlinear control architecture is proposed. Performance and stability of cooperative teleoperation systems are addressed under dynamic interactions between slave robots in the presence of model uncertainty. The robustness of the controller with respect to communication latency is also analyzed. </p> <p> Simulation and experimental studies demonstrate that the proposed approach is highly effective in all phases of a teleoperation task, i.e. in free motion and in contact with both flexible and rigid environments. </p> <p> The second approach involves finite-dimensional state-space models that incorporate the delay for free motion/ soft contact as well as rigid contact modes of operation. Local dynamic linearization control laws are employed to linearize robotic manipulators' dynamics. Model-based discrete-time Linear Quadratic Gaussian (LQG) controllers are proposed that can deliver a stable transparent response for each phase of operation. The robustness of these controllers with respect to parameter uncertainty is examined via the Nyquist analysis. Simulation results demonstrate the effectiveness of the proposed approach. </p> / Thesis / Master of Applied Science (MASc)

Identiferoai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/21907
Date01 1900
CreatorsSetoodeh, Peyman
ContributorsSirouspour, Shahin, Electrical and Computer Engineering
Source SetsMcMaster University
LanguageEnglish
Detected LanguageEnglish

Page generated in 0.0019 seconds