PlateGo and vAuto AI Team Lead

Worked on a novel acoustic sensor for self-driving vehicles to enable live full acoustic imaging of all surrounding acoustic sound events through sound events detection and localization (aka SELD) using microphone arrays and spatial audio. Built a novel method and problem definition and representation that enabled full simultaneous localization and detection of multiple objects of different classes from arbitrary directions, with an optimized and lightweight architecture that was capable of working live even on a RPi.Built an efficient annotated data acquisition and generation, both real and synthetic, using very efficient, cheap, and innovative setups.

Got introduced to internal code base and ecosystem. Introduced to Kaldi. Reproduced results of internal state-of-the-art Arabic automatic speech recognition (ASR), and optical character recognition (OCR).Didn’t get the chance to make much of an impact, but learned a lot about audio and OCR.

AI for autonomous vehicles. Focused on end-to-end driving imitation learning through visual perception, and transfer learning of driving experience from simulation to reality through domain abstraction. And deployed on a 1:10 RC car with edge compute NVIDIA Jetson TX2.Thesis title “SIM-TO-REAL CONDITIONAL END-TO-END SELF-DRIVING VEHICLETHROUGH VISUAL PERCEPTION”, supervised by Dr. Ibrahim Sobh.Demo videos:1- https://youtu.be/nAfe3akvviw?si=Pb-r4ZmVe143c17K2- https://youtu.be/nX1o0fayBts?si=xtJ6sOMEajeKn02a