Amiram Moshaiov

Interactive Evolutionary Multiobjective Search and Optimization of Set-Based Concepts

This paper deals with interactive concept-based multiobjective problems (IC-MOPs) and their solution by an evolutionary computation approach. The presented methodology is motivated by the need to support engineers during the conceptual design stage. IC-MOPs are based on a nontraditional concept-based approach to search and optimization. It involves conceptual solutions, which are represented by sets of particular solutions, with each concept having a one-to-many relation with the objective space. Such a set-based concept representation is most suitable for human-computer interaction. Here, a fundamental type of IC-MOPs, namely, the Pareto-directed one, is formally defined, and its solution is presented. Next, a new interactive concept-based multiobjective evolutionary algorithm is introduced, and measures to assess its resulting fronts are devised. Finally, the proposed approach and the suggested search algorithm are studied using both academic test functions and an engineering problem.

Set-based concept selection in multi-objective problems: optimality versus variability approach

This paper presents a novel approach to support the selection of conceptual solutions to multi-objective problems. The proposed method involves a comparison between concepts, based on the performances of sets of solutions that represent them. The set-based comparison of concepts is consistent with the so-called Toyota set-based concurrent engineering process. Such an approach discourages early exploitation of solutions and promotes extended exploration of the design space by means of sets of solutions. Both optimality and variability of concepts are considered, and their measures are devised to pose the selection problem as an auxiliary multi-objective problem. The auxiliary objectives are to maximise optimality and to maximise the variability. This highlights the inherent multi-objectivity of concept selection and supports decision-making under the possible contradictory nature of optimality and variability of concepts. Both academic and engineering problems are used to demonstrate the approach and to expose the inherent subjectivity of the measures, which are dependent on the selection of a window of interest by the decision-makers.

MOEA-Based approach to delayed decisions for robust conceptual design

This paper presents a modification of our multi-objective concept-based EC method to enhance the simultaneous development of a robust conceptual front. It involves a hierarchy of abstractive descriptions of the design, and a structured EC approach. The robustness, treated here by a novel MOEA approach, is for uncertainties, which result from delaying decisions during the conceptual design stage. The suggested procedure involves a robust non-dominancy sorting algorithm.

Towards a general tool for mechatronic design

This paper is a report on our initial attempt to develop a general tool and explore a non-traditional approach to mechatronic design, which emphasizes simultaneous mechanical and control design. The approach is centered around our novel GA-based design methodology that is specially developed for exploring possible design alternatives in early design stages with a particular emphasis on interdisciplinary design. The suggested methodology, as applied to mechatronics, transforms the common GA-based search in the space of individual solutions into a search that relies heavily on exploration of the space of mechatronic sub-concepts. In addition to demonstrating the validity of our proposed technique to mechatronics, we show the relevance of our methodology to control design at large.

Multi-objective evolution of robot neuro-controllers

This paper concerns a non-traditional evolutionary robotics approach to robot navigation. Navigation is presented as a problem of two conflicting objectives. The first concerns a classical “amalgamated” objective, which has been traditionally used to increase speed, move straight as possible, and at the same time avoid obstacles. The second objective is devised to simultaneously encourage a sequential acquisition of targets. To solve the presented problem a modification of the well known NSGA-II algorithm has been performed. The proposed approach is tested using a simulation of a Khepera. The study sheds light on different aspects of the aforementioned problem and on the applicability of evolutionary multi-objective optimization to the simultaneous learning of a variety of controllers for deferent behaviors. Finally, based on this initial study, future work is suggested, which may allow to shift such multiobjective evolutionary studies from toy problems to more realistic situations.

Concept-based multi-objective problems and their solution by EC

Recent studies on the support of engineers during conceptual design resulted in a non-traditional type of Multi-Objective Problems (MOPs), namely concept-based ones. In concept-based MOPs the focus is on conceptual solutions that are represented by sets of particular solutions. This means that a concept has a one-to-many relation with the objective space. Such a set-based concept representation is most suitable for human-computer interaction. In concept-based MOPs concept-related preferences could be easily incorporated with or without range-related preferences. This paper provides an overview of studies on concept-based problems, which have been conducted at Tel-Aviv University, and suggests some future research directions.

Set-based concept selection in multi-objective problems involving delayed decisions

This paper introduces a computational approach to support concept selection in multi-objective design. It is motivated by: (1) a common need to delay some decisions during conceptual design due to the presence of uncertainties; and (2) intentional delay of decisions for the purpose of maintaining several optional concepts, as suggested by the concurrent engineering procedure of Toyota. Here, for the first time, a multi-objective set-based concept (SBC) selection problem with delayed decisions is formulated and solved. SBCs are conceptual solutions, which are represented by sets of particular solutions, with each concept having a one-to-many relation with the objective space. Several novel notions, such as higher-level concepts, multi-model concepts and robust concepts to delayed decisions, are defined and used. These lead to an auxiliary multi-objective decision problem. The auxiliary objectives are concept optimality and variability, both paramount to concept selection, with concept variability strongly supporting the idea of intentionally keeping several useful alternatives as long as possible. Academic and engineering examples are provided to demonstrate the proposed approach and its applicability to real-life problems. The results demonstrate that the suggested technique may well support the process of delayed decision either when needed or when deliberately done.

Concept-based IEC for Multi-objective Search with Robustness to Human Preference Uncertainty

The recently introduced concept-based IEC approach is extended to include robustness to uncertainties of the preferences, of the decision makers, towards sub-concepts and concepts. A novel measure is presented, which allows the decision makers to assess the relative robustness of concepts in a multi-objective evolutionary search process. Examples of interactive scenarios are given to demonstrate the proposed approach.

Concept-based interactive evolutionary computation for multi-objective path planning

In this paper we introduce a new methodology for interactive path planning. It is based on our interactive concept-based evolutionary computation technique, which has been originally developed for engineering design. The planning is divided into two levels. At the lower level paths are planned based on models, which are used by the computer to evaluate different path alternatives. At the higher level, which is termed conceptual level, solution alternatives are described by words and interrogated by a human-machine interaction. A structured interactive evolutionary computation is implemented to search for candidate paths by a Pareto-directed methodology. Examples are employed to demonstrate the proposed approach

Dynamic-Escape-Zone to Avoid Energy-Bleeding Coasting Missile

This study deals with the problem of planar safe flight of a fighter aircraft in the presence of a pursuing “energy-bleeding” coasting missile. A“dynamic-escape zone” is proposed as a novel real-time aid for the aircrew to decide when to commence the latest yet safe-kinetic-evasion (that is, missile outrunning) from an already launched missile. Such safe evasion is based on exhausting the energy of the adversary missile to avoid missile rendezvous. At the instance of its calculation, the dynamic-escape zone defines a set of states along the current line of sight. When the evading aircraft is located inside the dynamic-escape zone, survivability can be guaranteed by applying optimal kinetic-evasion strategies, which result in outrunning the missile. A method is suggested to calculate the dynamic-escape zone in real-time for online applications. The calculation is based on several conservative assumptions, as well as on knowing the missile and aircraft parameters. Simulations are used to demonstrate the eff...