Thesis (M.A.)--University of Hong Kong, 2002. / Includes bibliographical references (leaves 103-105).

Thesis (Ph. D.)--University of Hong Kong, 2002. / Includes bibliographical references (leaves 379-401).

Thesis (Ph. D.)--University of Hong Kong, 2002. / Includes bibliographical references (leaves 354-392).

Thesis (M.S. in Computer Science)--Naval Postgraduate School, March 2003. / Thesis advisor(s): Neil C. Rowe. Includes bibliographical references (p. 65-67). Also available online.

published_or_final_version / Mechanical Engineering / Doctoral / Doctor of Philosophy

Not available / text

The concept of trust in multi-agent systems (MASs) has received significant attention in recent years, and a number of approaches have been proposed to enable agents to form, maintain and use trust relationships in their dealings with others.  However, current approaches do not adequately address <i>highly dynamic </i>multi-agent systems, where the population and structure changes frequently.  For example, agents may frequently join and leave, and ad-hoc structures may form in response to emerging situations. In these highly unstable environments, trust can be difficult or impossible to build with existing techniques.  Trust matters most when risk is involved, but in situations of extreme uncertainty, the risk may be too great to permit any interactions, resulting in a breakdown of the system. In this thesis, we propose a general approach for trust evaluation and decision-making in highly dynamic multi-agent systems.  First, we present a model of <i>stereotypes, </i>which allows agents to build tentative trust relationships with others on the basis of visible features.  We show that this approach can help agents to form trust relationships, despite a high degree of social dynamicity.  We present a method of selecting providers of trust evidence, when those providers may be stereotypically biased. Secondly, we present a trust decision-making model which employs <i>controls</i>, as well as trust evaluations and stereotypes, in order to facilitate initial interactions when trust is low or absent, and bootstrap dynamic societies.  We show that control can be used initially to enable interactions.  As trust builds, control can be reduced.  Our approach is general and applicable to existing models of trust in MASs.  We evaluate our model within a simulated multi-agent environment characterised by high degrees of dynamicity and structural change.

published_or_final_version / Electrical and Electronic Engineering / Master / Master of Philosophy

To improve the accuracy in capturing the motion of deformable objects, a reconfigurable multi-camera active-vision system which can dynamically reposition its cameras online is proposed, and a design for such a system, along with a methodology to select the near-optimal positions and orientations of the set of cameras, is presented. The active-vision system accounts for the deformation of the object-of-interest by tracking triangulated vertices in order to predict the shape of the object at subsequent demand instants. It then selects a system configuration that minimizes the expected error in the recovered position of each of these vertices. Extensive simulations and experiments have verified that using the proposed reconfigurable system to both translate and rotate cameras to near-optimal poses is tangibly superior to using cameras which are either static, or can only rotate, in minimizing the error in recovered vertex positions.

To improve the accuracy in capturing the motion of deformable objects, a reconfigurable multi-camera active-vision system which can dynamically reposition its cameras online is proposed, and a design for such a system, along with a methodology to select the near-optimal positions and orientations of the set of cameras, is presented. The active-vision system accounts for the deformation of the object-of-interest by tracking triangulated vertices in order to predict the shape of the object at subsequent demand instants. It then selects a system configuration that minimizes the expected error in the recovered position of each of these vertices. Extensive simulations and experiments have verified that using the proposed reconfigurable system to both translate and rotate cameras to near-optimal poses is tangibly superior to using cameras which are either static, or can only rotate, in minimizing the error in recovered vertex positions.