Despite the rapid development of technology, there is saturation in the increase in the performance of the computing hardware, which requires new approaches to calculation. The work explores the utilization of Heterogeneous System on Chip (HSoC) technology and the stereo vision algorithm implementation. The following topics are examined: heterogeneous embedded system development approaches, combining abstraction levels, effective internal communication, co-processing using different computing hardware, real-time processing, implications for modern AI-based algorithms and realization of an efficient computer vision pipeline.
About the author
Rihards Novickis was born in 1993 in Daugavpils. He has received a Bachelor’s degree in Electronics in 2014 and a Master’s degree in Electronics in 2016. He has been an employee of Belam-Riga un SAF Tehnika companies. Since 2016, he has been an electronics engineer, later a scientific assistant at the Institute of Electronics and Computer Science , where he is currently a researcher. Since 2019, he has been an educational assistant at Riga Technical University. Rihards has received an appreciation award for one of the most important achievements in Latvian science in 2018 for “Original approach for neural network architecture implementation in field-programmable gate arrays” and has participated in the European Commission’s “System-on-a-Chip” Working Group as Latvia’s National SoC expert. His scientific interests include heterogeneous SoCs and their application in improving algorithm performance and energy efficiency.