Masoud Daneshtalab

Mälardalen University, KTH Royal Institute of Technology, University of Turku

Papers

Total Citations

H-Index

About

Masoud Daneshtalab is a leading researcher in embedded systems, reconfigurable computing, and autonomous vision architectures. His work bridges the gap between hardware acceleration and intelligent perception, particularly for real-time applications in robotics and self-driving cars. He pioneered compact convolutional neural network designs for embedded stereo vision systems, enabling efficient depth, luminance, and shape extraction on resource-constrained platforms—a contribution cited over 15 times in the autonomous systems community. Daneshtalab also advanced optimization theory by developing multi-population parallel imperialist competitive algorithms for solving systems of nonlinear equations, with applications spanning economics, engineering, and mechanics. His innovative framework, FIST, introduced a method to interleave spiking neural networks on Coarse Grained Reconfigurable Architectures (CGRAs), allowing cognitive embedded systems to simultaneously process auditory and visual data. Further demonstrating his versatility, he accelerated stereo vision algorithms using SSE3, AVX2, and CUDA, dramatically reducing execution time for real-time depth detection. Daneshtalab’s work is foundational for next-generation autonomous systems, where low-latency, high-efficiency perception is critical.