Pedro Nuno De Melo Antunes

Hardware Architecture, FPGA Acceleration, Space Applications, Onboard Inference, Neural Network Accelerators, Bit-Serial Computing, High-Level Synthesis (HLS), Vitis AI, Embedded Systems.

Pedro Antunes is a researcher specializing in computer hardware architecture and embedded systems, with a core focus on enabling advanced artificial intelligence in extreme environments. His research addresses the critical computational, energy, and data constraints of modern space missions by developing high-performance, radiation-tolerant onboard computing systems.

Pedro's work spans the comprehensive lifecycle of hardware acceleration for aerospace applications. His contributions include publishing foundational surveys on FPGA-based neural network accelerators for space, evaluating deployment constraints across realistic space use cases using AMD/Xilinx hardware (such as Vitis AI and Vitis HLS), and pioneering custom hardware architectures. Notably, his work on bit-serial matrix multiplication accelerators (such as bitSMM) introduces runtime-configurable precision and hardware resiliency mechanisms designed to maximize energy efficiency and fault tolerance. Through onboard inference, data reduction, and event detection, his research aims to alleviate downlink bandwidth pressures and drive the next generation of autonomous spacecraft.