Wyniki wyszukiwania - BazTech

1

A spiking neural network based on thalamo-cortical neurons for self-learning agent applications

Huderek Damian, Szczęsny Szymon, Pietrzak Paweł, Rato Raul, Przyborowski Łukasz

International Journal of Applied Mathematics and Computer Science

|

2024

|

Vol. 34, no. 3

467--483

EN

The paper proposes a non-iterative training algorithm for a power efficient SNN classifier for applications in self-learning systems. The approach uses mechanisms of preprocessing of signals from sensory neurons typical of a thalamus in a diencephalon. The algorithm concept is based on a cusp catastrophe model and on training by routing. The algorithm guarantees a zero dispersion of connection weight values across the entire network, which is particularly important in the case of hardware implementation based on programmable logic devices. Due to non-iterative mechanisms inspired by training methods for associative memories, the approach makes it possible to estimate the capacity of the network and required hardware resources. The trained network shows resistance to the phenomenon of catastrophic forgetting. Low complexity of the algorithm makes in-situ hardware training possible without using power-hungry accelerators. The paper compares the complexities of hardware implementations of the algorithm with the classic STDP and conversion methods. The basic application of the algorithm is an autonomous agent equipped with a vision system and based on a classic FPGA device.

2

Parallel cross-section recognition of geometrical features for selected machine parts

Golaszewski Marcin, Krolikowski Marcin, Powalka Bartosz

Journal of Machine Engineering

|

2021

|

Vol. 21, No. 3

60--79

EN

Increase of automation and autonomy of production is the latest trend incorporated into Industry 4.0 objectives. Production autonomy is very desirable in the field of damaged parts replacement. To fulfill this goal numerous reverse engineering systems have been developed that support geometry recognition from the 3D scan data. This study is focused on converting non-parametric geometry representation of shaft-type elements into a CAD model with a rebuilt feature tree. Algorithms are based on the analysis of parallel cross-sections. The proposed system is also capable of identification of additional geometric features typical for 2.5 axes milling such as pockets, islands and outer walls. The proposed algorithms are optimized to increase efficiency of the process. Initial identification parameters are selected with respect to defined criteria, e.g., identification accuracy, computing power and scanning accuracy. Described algorithms can be implemented in reverse engineering systems.

3

Navigation system without GPS

Typiak A., Typiak R.

Logistyka

|

2009

|

nr 6

CD-CD

EN

This paper focuses on the latest technology trends for navigating in difficult urban, indoor, and underground environments where typical Global Positioning Systems (GPS) fail to work correctly. The latest alternative navigation technologies based on opto-elektronical techniques will be described. Presented technologies include ladar aided inertial navigation systems (INS). Tightly integrating this technologies should lead to navigation performance similar to that achieved in today’s GPS/INS integrated systems.

PL

Referat ten skupia się na nowoczesnych trendach w dziedzinie nawigacji w skomplikowanym miejskim, bądź podziemnym środowisku, gdzie zastosowanie układów GPS nie jest moSliwe. Praca zawiera opis najnowszych alternatywnych metod nawigacji opartych na technice opto-elektronicznej. Zaprezentowane rozwiązania obejmują wykorzystanie inercyjnego systemu nawigacji (INS) opartego na ladarze. Zintegrowanie tych technologii ze sobą powinno spowodować osiągnięcie wydajności nawigacyjnej podobnej do tej, którą otrzymuje się wykorzystując zintegrowane systemu GPS/INS