Defense Session of the Doctoral Thesis No. 52 of PPGEE - Analysis and construction of hardware accelerators for calculating the shortest path in real-time robot route planning

STUDENT: Linton Thiago Costa Esteves

 

DATE: 10/30/2024

 

TIME: 08:00

 

LOCATION:  https://conferenciaweb.rnp.br/sala/wagner-luiz-alves-de-oliveira

 

TITLE: Analysis and construction of hardware accelerators for calculating the shortest path in real-time robot route planning

 

KEYWORDS: Dijkstra; Shortest Path; PRM; FPGA; Robotics;

 

ABSTRACT:

 

This work proposes an analysis and optimization for calculating the shortest path in route planning for mobile robots. The proposed solution aims to present a high-performance alternative that can meet the time constraints required for real-time processing for robots. For this, an architecture focused on parallelism was built to be embedded in dedicated hardware. By exploiting parallelism, the solution aims to present, in addition to a performance improvement, a dynamic adaptation to changes in the graph of possible movements to be analyzed, since edges can be inserted or removed in a temporally random manner according to changes in the environment. This work demonstrates the developed architecture together with its results. The application graph is updated efficiently through an obstacle matrix, resulting in a notable improvement of 120 times for graphs with 1,024 nodes. By using a low-cost device such as the Cyclone IV E, performance is approximately 20 times higher than that of an equivalent software application.

 

 

MEMBERS OF THE BOARD:

 

Prof. Dr. Wagner Luiz Alves de Oliveira (UFBA - Advisor)

Prof. Dr. Paulo César Machado de Abreu Farias (UFBA - Advisor)

Prof. Dr. Tiago Trindade Ribeiro (UFBA) Prof. Dr. Anfranserai Morais Dias (UEFS)

Prof. Dr. Edward David Moreno Ordonez (UFS)

Prof. Dr. Nelson Alves Ferreira Neto (PPGEE Graduate)

 

On 10/22/2024

Em 24/10/2024