Smart Shoe Performing Real Time Motion Classification using Machine Learning for External Motion Alerts in Vehicles.

Presenters

Kaleb KeyFollow

Disciplines

Other Computer Engineering

Abstract (300 words maximum)

As networked sensors become more ubiquitous and integrated into the Internet-of-Things, various technologies can be leveraged to provide an increased amount of safety to the public. In this research we have constructed a pressure sensor within a shoe covering two regions of detection: heel and toe. The pressure sensor is centered around polymeric foil Velostat which has the property of changing its resistance dependent upon amount of pressure applied. When made into pads placed inside of the shoe and connected in a circuit, we can record the pressure values from the pads over time and use them to train decision trees for classification. We assess the performance of our sensor + model in a deployment scenario where inferencing is accomplished at the user wearable. We envision our wearable to assist in alerting approaching vehicles of the type of mobility of surrounding users with our system.

Academic department under which the project should be listed

SPCEET - Electrical and Computer Engineering

Primary Investigator (PI) Name

Billy Kihei

This document is currently not available here.

Share

COinS
 

Smart Shoe Performing Real Time Motion Classification using Machine Learning for External Motion Alerts in Vehicles.

As networked sensors become more ubiquitous and integrated into the Internet-of-Things, various technologies can be leveraged to provide an increased amount of safety to the public. In this research we have constructed a pressure sensor within a shoe covering two regions of detection: heel and toe. The pressure sensor is centered around polymeric foil Velostat which has the property of changing its resistance dependent upon amount of pressure applied. When made into pads placed inside of the shoe and connected in a circuit, we can record the pressure values from the pads over time and use them to train decision trees for classification. We assess the performance of our sensor + model in a deployment scenario where inferencing is accomplished at the user wearable. We envision our wearable to assist in alerting approaching vehicles of the type of mobility of surrounding users with our system.