README

This repository contains the code and necessary files to build and run the Pedestrian Audio Wearable System (PAWS) smartphone applications.

PAWS is a low-power connected system for improving pedestrian safety in current and future smart cities. PAWS uses microphones embedded into a headset combined with low-power feature extraction, signal processing, and machine learning, for detecting, localizing, and providing alerts of oncoming vehicles to pedestrians in noisy environments.

• Note: The PAWS project was at some point known as SEUS, which is why SEUS is used at some points rather than PAWS.

The contents of this repository are summarized below.

• PAWS: Contains the source code and necessary files for the Pedestrian Audio Wearable System (PAWS) smartphone application.
  • Application: Contains the Android project for building and loading the PAWS smartphone application.
  • SEUS: This folder contains the PAWS car detection and localization classifier models. To run the PAWS smartphone application, this folder must be placed into the Downloads folder of the smartphone.
• PAWS_LE: Contains the source code and necessary files for the Pedestrian Audio Wearable System Low-Energy (PAWS LE) smartphone application. PAWS LE is the low-energy variant of PAWS and has finer granularity car localization, achieved by using a technique called Angle via Polygonal Regression (AvPR).
  • Application: Contains the Android project for building and loading the PAWS LE smartphone application.
  • SEUS: This folder contains the PAWS LE car detection and localization classifier models. To run the PAWS LE smartphone application, this folder must be placed into the Downloads folder of the smartphone.
• PAWS_Features: Contains the Android project for the test application that displays the raw features obtained from the PAWS/PAWS_LE headset.
• PAWS_Wireless_Latency: Contains the Android project for the test application that measures the wireless latency between the PAWS/PAWS_LE headset using a button connected to the smartphone's audio jack.

Reproducing the PAWS system

To reproduce the PAWS system, you must first install the PAWS smartphone application; details on how to do this can be found in the PAWS/Application/ folder.

Next, you must reproduce the PAWS front-end PCB. Details on how to do this can be found in the PAWS-FrontEnd repository.

Using the PAWS Smartphone Application

A manual detailing how to use the smartphone application can be found in PAWS/PAWS_Manual.pdf. This file details basic app usage and how to train your own car detection models.

Reproducing the PAWS LE System

To reproduce the PAWS LE system, you must first install the PAWS LE smartphone application; details on how to do this can be found in the PAWS_LE/Application/ folder.

Next, you must reproduce the PAWS LE front-end PCB. Details on how to do this can be found in the PAWS-FrontEnd repository.

Using the PAWS Smartphone Application

A manual detailing how to use the smartphone application can be found in PAWS_LE/PAWSLE_Manual.pdf. This file details basic app usage and how to train your own car detection models.

More About the PAWS Project

To learn more about PAWS, please visit our project page, or contact us at: stephen.xia@columbia.edu.

This repository is part of the Pedestrian Audio Wearable System (PAWS) project of the Intelligent and Connected Systems Lab (ICSL), Columbia University. For more information about our latest projects, please visit our group website.