The research in scientific visualization addresses the increasing reliance on modern visualization techniques to steer decision-making processes in the design and
virtual prototyping of complex systems.
The primary research issues involve the development of visualization techniques that can be used effectively in collaborative design environments.
These environments are to be used to make multiobjective decisions under uncertainty and to enable rapid trade-off decisions by geographically distributed participants in the context of Rapid Virtual Prototyping.
Visualization in Multiobjective Design (Lewis, Bloebaum, Winer)
Our goal is to provide effective decision support in the form of interactive visual tools in
multiobjective design optimization problems. These tools couple both the design spaces and the
performance spaces using novel mapping techniques and provide the designer(s) with trade-off
information. These techniques are also being built into Internet-based visualization systems for
use in distributed design processes.
Visual Design Steering (VDS; Lewis, Bloebaum, Winer)
Visualization is used in real time to allow the designer to 'steer' the design to a more
accurate and efficient solution. Several projects are under way to develop VDS methods and tools to
enable better trade-off decisions in complex analysis, to determine the best way to schedule design
tasks of large-scale systems, and to provide mechanisms for real-time design decision making in
collaborative environments.
Medical API (Bloebaum, Winer)
This project, under way in partnership with the
Toshiba Stroke Center and SGI, involves the development of a multipurpose high-end medical image analysis and
visualization programming interface.
OPTIMIZATION IN DESIGN
The optimization research spans a number of primary areas in the development of design tools to model, solve, and optimize a system. These areas include developing optimization approaches that capitalize on
parallel or distributed processing capabilities, developing heuristic and memetic algorithms for use with discrete and complex problems, and techniques for multiobjective optimization formulation and solution.
Multiobjective Optimization in Design (Lewis, Bloebaum)
Our goal is to model multiobjective design optimization problems, both discrete and continuous, and to develop
methods to solve them based on decision theory and using the stated preferences of the decision maker. Using the
actual preferences precludes the use of assumed weights, preference strengths, or scales. Visualization techniques
are also developed to aid designers in making effective decisions in n-dimensional problems.
Memetic Algorithms in Parallel Environments (Lewis, Bloebaum, Mayne)
Our goal is to develop memetic, or hybrid, techniques to solve complex optimization problems in parallel
environments. The technique combines a heuristic solver, a genetic algorithm, with gradient-based searches, and
uses novel switching techniques to change from one technique to another based on design space
characteristics.
New Optimization Algorithms for Massively Parallel Implementation (Bloebaum)
New optimization algorithms are being developed to take best advantage of the parallel and distributed computing
infrastructures now readily available in most industries and academic environments. The goal is to develop
algorithms that are scalable with availability of processors, so as to reduce efficiency loss due to message
passing. Examples include a hybrid genetic simulated annealing method, a direct parallel optimization method, and a
concurrent subspace optimization method.
Optimization and Design in a PC Windows Environment (Mayne)
The goal here is to explore optimization tools and other design calculations using Visual C++ with the intention of developing
computational techniques and programs ready for use in a PC Windows environment. Optimization methods are being
considered for implementation in a convenient C/C++ format, and visualization methods for monitoring and
understanding optimum design solutions are being explored. Emphasis is on the development of practical tools for
use in mechanical design applications.
VIRTUAL REALITY/HAPTICS
Virtual reality is fast emerging as an indispensable tool for solving a wide variety of engineering problems, such as manufacturing, biomedical devices, virtual prototyping, scientific visualization, and
transportation. Here at UB, the research effort is focused on integrating such diverse technologies as haptics, real-time hardware/human-in-the-loop simulations, and 3-D visualization systems.
Adverse Condition Alerting Systems (Singh, Kesavadas, Mayne)
This work focuses on the development of algorithms to provide cues to drivers to assist them in preventing
spin-outs in inclement conditions. A virtual reality-based driving simulator is used to test the human-in-the-loop
system.
User-customized Telerehabilitation Environment (Krovi)
Our research focuses on the development of a low-cost haptically-enabled virtual driving environment and
a series of exercises/protocols to serve as an integrated low-cost diagnostic and therapeutic tool for both assessment of UL dysfunction
and UL motor rehabilitation. The VE driving paradigm explored over here, offers a promising and cost-effective method for
objective/quantitative assessment of UL performance while performing both unilateral and bimanual sensorimotor tasks in the context of
one higher activities of daily living (AsDL).
The Smart Car Project - A Case Study of Computer-Mediated Interfaces (Krovi)
In this research, we investigate the development, implementation and testing of an inexpensive
scaled-prototype "Smart Car" test bed equipped with a real-time mediated control system. This test bed enables us to study several
concepts including: (i) Mediation of human user control of complex robot systems; (ii) Multi-user shared teleoperation; and (iii)
Robustness of the control in the presence of varying grades of communication that are critical to a number of current and future
generations of military/civilian systems
Driving Simulation Exposure for Phobia Victims (Winer)
This project involves developing a driving simulator model to enable victims of accidents to be exposed to driving virtually as a first
step to driving a vehicle once again. This project is being implemented in partnership with the UB Department of Psychology.
DESIGN THEORY
The design theory research focuses on decision making in the design of large-scale systems marked by distributed and collaborative design processes. Research activities include modeling groups of decision makers and
simulating their impact on the final product design. Additional topics include modeling and simulation of system uncertainty in design decision making, and developing a decision-based design framework for flexible
systems.
Application of Game Theoretical Principles to the Design of Large-Scale Systems (Lewis)
Our goal is to model a complex design process as a series of games among designers or
players. Depending upon how the designers make their decisions, or react to other decisions, the outcome of the
game—the end product—changes. Focus is on modeling the inherent trade-offs and scientific
principles.
Decision-Based Design Framework for Flexible Systems (Lewis)
Our goal is to develop a consistent decision-making framework based on concepts from operations research,
economics, marketing, and product design. The objective of the decision-making framework is to maximize net present
value of any product design project. Design engineering decisions are propagated through to profit, giving
engineers a sense of how their decisions affect not only the product performance, but also company
performance.
Distributed Design: Convergence and Stability (Lewis, Bloebaum)
Our goal is to study the convergence and stability properties of distributed, but coupled, design
processes. Conditions for convergence and stability under various conditions and design scenarios are
developed and studied.
Uncertainty Modeling in Design (Lewis, Bloebaum)
Our goal is to study the uncertainty in a design process that arises because of the coupling among
designers. This uncertainty occurs when one designer does not have information needed from another
designer. Sensitivities, trade-offs, and rational decision making are used to study the effects of this
uncertainty.
MECHATRONICS
Mechatronics—a blend of mechanics, electronics, information technology, and computers—has come to embody an integrated approach for the design, analysis, and implementation of a range
of complex engineering systems. Research activities include cooperating systems of robots, haptics, and disk drives, with an emphasis on examining the trade-offs between functionality implemented in hardware or software, with an
underlying emphasis on developing reliable and robust systems.
Cooperative Payload Transport by Robot Collectives (Krovi)
Cooperative material-handling by a fleet of decentralized manipulation agents has many applications ranging from
hazardous waste removal, material handling on the shop floor, to robot work crews for planetary colonization. Our long-term goal is the
development of a theoretical and operational framework to model, analyze, implement and validate cooperative payload transport
capabilities in such distributed robot collectives.
Rapid Virtual and Physical Prototyping of Electromechanical Systems (Krovi)
The goal of this work is to ratify the paradigm for rapid development, refinement, and implementation of both
novel electromechanical/mechatronic designs and effective real-time control systems by combining methods from extensive simulation-based
virtual testing with rapid human-in-the-loop and hardware-in-the-loop physical testing.
University at Buffalo: Mechanical and Aerospace Engineering
