作者: Abhishek Nayak , Sivakumar Rathinam , Swaminathan Gopalswamy
DOI:
关键词:
摘要: Project Description: The objective of this project is to explore how an autonomous vehicle must identify and safely respond to emergency vehicles using sound, vision and other onboard sensors. Emergency vehicles can belong to police, fire, hospital and other responders. The autonomous vehicle in the presence of an emergency vehicle must have the ability to accurately sense its surroundings in real-time and be able to safely yield to the emergency vehicles. This project identifies the presence of emergency vehicle mainly through onboard vision and safely maneuvers the vehicle and stops it. The project also explored using onboard sound sensors to identify the emergency vehicles. Several proof of concept demonstrations were performed, and the results were presented in several conferences and in poster sessions. The project also lead to the publication of a conference article and fully/partly supported the thesis of an MS and several Ph. D. students. Data Scope: Audio Dataset: The dataset used to train the machine learning algorithms for audio based EV detection include a large collection of audio files pertaining to emergency and non-emergency vehicles as well as other road sounds. The audio dataset is annotated into 527 classes representative of various sounds. The data is processed to extract audio files relevant to emergency and non-emergency classes by extracting features like zero crossing rate, spectral parameters, chromatograms etc., using the PyAudio Analysis library. Emergency Vehicle videos: Videos of Emergency vehicles collected from inside the Autonomous vehicle using BFLY-PGE-20E4C-CS 1/1.8" Blackfly® PoE …