O grupo de métodos de otimização do programa de pós-graduação em pesquisa operacional do ITA e da Unifesp está recebendo, com apoio da Fapesp, a visita do professor Alf Kimms, da Universität Duisburg-Essen, Lehrstuhl für Logistik und Operations Research. Dentre as atividades programadas, teremos duas palestras abaixo especificadas. Estão todos convidados.

DISASTER MANAGEMENT SUPPORT BY OPERATIONS RESEARCH

PALESTRANTE: Prof. Dr. Alf Kimms (Universität Duisburg-Essen, Lehrstuhl für Logistik und Operations Research)

DATA E HORÁRIO: 14/03/2024 (quinta-feira), às 9h00.

LOCAL: Auditório Celso Renna, localizado na Divisão de Ciência da Computação (IEC) (ao lado da secretaria da IEC)

INFORMAÇÕES: Profa Dra Mariá C. V. Nascimento (Este endereço de email está sendo protegido de spambots. Você precisa do JavaScript ativado para vê-lo.), IEC

Abstract

Since two decades we have been working on operations research

models and methods to support the management of disasters. Five

applications will briefly be described in this talk:

(1) Task Force Deployment

Imagine that for a big event ambulances must be positioned in such a

way that help can be provided fast. Questions like what is the minimal

number of ambulances needed and where should they be located arise.

(2) Evacuation Planning in Urban Areas

Imagine that for some reason you need to evacuate people out of an

urban area. How should vehicles be routed through the streets such

that the evacuation is completed fast?

(3) Road Clearance and Relief Goods Delivery

Imagine that for some reason the road infrastructure is partially de-

stroyed and areas in need are isolated. Which parts of the infrastruc-

ture should be repaired first so that aid can be provided?

(4) Fighting Fires and Floods

Imagine that a flood or a fire threatens a certain area. Where should

you send helping troops to keep things under control?

(5) Drone Trajectory Planning with Obstacles

Imagine that a drone can brought into a certain area by a vehicle.

Where should you launch the drone how should that drone fly towards

a destination point so that obstacles are not hit?

For these applications mainly basic modeling ideas are described

and by far not all technical details are revealed. Details are given for

two selected subproblems which might be relevant in other applica-

tions as well: (a) a linear approximation for computing the Euclidean

length of a vector and (b) a linear formulation for testing if two convex

polygons overlap. The former one for instance is important to model

moving objects that may change their direction. The latter one is

important for instance in the context of cutting and packing irregular

objects.

EXPLOITING THE PRINCIPLE OF LOCAL VALUE FOR MODELING ORDERED OBJECTS

PALESTRANTE: Prof. Dr. Alf Kimms (Universität Duisburg-Essen, Lehrstuhl für Logistik und Operations Research)

DATA E HORÁRIO: 15/03/2024 (sexta-feira), às 9h00.

LOCAL: Laboratório MEMO, sala 27, Parque Tecnológico UNIFESP

INFORMAÇÕES: Prof Dr Felipe Bueno (Este endereço de email está sendo protegido de spambots. Você precisa do JavaScript ativado para vê-lo. ).

LINK PARA PARTICIPAÇÃO REMOTA: https://meet.google.com/vkj-roff-jqs

Abstract

Imagine that your application requires to consider sequences of ob-

jects. Examples of this kind for instance occur at container terminals

where stacks of containers are stored and you need to know in what

sequence these containers are put on each other in order to retrieve

a particular one. Another example are shunting yards where railroad

carriages are parked. When you need to put a train together, you need

to know which carriages are placed on what tracks and in which order

to select the right carriages for that particular train. In more general

terms, whenever you encounter a last–in–first–out (LIFO) storage sys-

tem, you have to keep track of the sequence of items.

Typical planning problems in such situations include the Block

Relocation Problem: In order to retrieve all items from LIFO stacks

in a particular order, relocations might be necessary to be able to

reach the items that are to be retrieved first. To goal is to minimize

the number of required relocations.

To model such ordered objects we present an idea that heavily

makes use of the so–called local value. A concept that is known by

every first grader, but that has not been used in this context. We will

demonstrate how relocations and retrievals from LIFO stacks can be

modeled in a novel way.

Seminário de Autoavaliação do PPG-PO ITA-Unifesp

28 de abril de 2022

9h - 9h30: Abertura: Prof. Elbert Macau - Presidente da Câmara de Pós-graduação e Pesquisa do ICT-Unifesp e Prof. Erico Rempel - Chefe da Divisão de Pós-graduação e Pesquisa do ITA

9h30 - 10h: Apresentação da Coordenação sobre o estágio atual do PPG-PO em relação ao planejamento estratégico realizado em 2019

10h-11h20: Palavra aberta

11h20-12h: Conclusões e encerramento