Banca de QUALIFICAÇÃO: LUÍS BRUNO PEREIRA DO NASCIMENTO

Uma banca de QUALIFICAÇÃO de DOUTORADO foi cadastrada pelo programa.
DISCENTE : LUÍS BRUNO PEREIRA DO NASCIMENTO
DATA : 06/07/2018
HORA: 09:00
LOCAL: Auditório do nPITI
TÍTULO:

Path Planning Based on Fast Propagation Probabilistic Foam Method Applied to Robotic Systems

PALAVRAS-CHAVES:

Autonomous Robots, Sampling-Based Planning, Probabilistic Foam, Bubbles,  Metrics in Workspace, A-Star Algorithm.

PÁGINAS: 107
GRANDE ÁREA: Engenharias
ÁREA: Engenharia Elétrica
SUBÁREA: Eletrônica Industrial, Sistemas e Controles Eletrônicos
ESPECIALIDADE: Automação Eletrônica de Processos Elétricos e Industriais
RESUMO:

Path planning is one of the main problems in autonomous robotics, since it is a process that provides an obstacle-free path for the robot to move from an initial configuration to a final configuration. Planning for mechanisms with many degrees of freedom requires considerable computational effort and sampling-based planning strategies have been extensively explored in these problems. Probabilistic Foam (PFM) is a sampling method that uses a structure called bubble that has the advantage of guaranting a free region for safe maneuverability. In PFM, a structure called probabilistic foam, constituted by a set of bubbles, propagates through the free space starting from an initial configuration to a final configuration, abling to extract an obstacle-free path.  Although PFM is a technique that solves the problem of planning, it has three main disadvantages. One of them is the need of computing distances in configuration space from robot to the obstacle region, which needs to be represented explicitly. This is impractical for applications in problems with many degrees of freedom. Another disadvantage is that the foam propagates faster through wider spaces on the map, rather than following regions that may lead to shorter paths. A third problem is the computation of an unnecessary number of bubbles, requiring excessive computational resources. In this context, a new strategy for the propagation of probabilistic foam, Fast Probabilistic Foam (FPF), inspired by the search algorithm A *, is proposed in order to reduce the density of the foam and search for shorter paths using a cost function for each bubble. In addition, a new bubble definition is proposed, based on workspace distance information, turning unnecessary the explicit computation of the obstacle region in configuration space and enabling the application of FPF to high dimensions robotic problems. Preliminary results were obtained by applying the improved planning method to an lower limb active orthosis, in tasks involing overcoming obstacles. Simulation results suggest that the new approach used in the probabilistic foam method presents better results in comparison with its previous approach.

MEMBROS DA BANCA:
Presidente - 1242315 - PABLO JAVIER ALSINA
Interno - 1153006 - LUIZ AFFONSO HENDERSON GUEDES DE OLIVEIRA
Externo ao Programa - 1445637 - WALLACE MOREIRA BESSA
Externo à Instituição - ARMANDO SANCA SANCA - UEFS