Evaluation of Multiple Approaches for Automated Gesture Recognition

Period of Performance: 07/01/2003 - 04/30/2004

$100K

Phase 1 STTR

Recipient Firm

Develosoft Corp.
1630 30th Street, Suite 121
Boulder, CO 80301
Firm POC
Principal Investigator

Research Institution

University of California, Santa Barbara
Office of Research
Santa Barbara, CA 93106
Institution POC

Abstract

DeveloSoft proposes to use its considerable carrier automation experience to investigate three main approaches for gesture recognition. First, we will evaluate either Measurand ShapeTape or micro-accelerometers for direct measurement of arm and hand positions. Second, we will investigate an imaging approach using LEDs or reflective tape... both with specially tuned optics and frequencies which work in sun glare, etc... Third, we propose to investigate automated recognition of voice commands through a special bone conductive microphone which works in 140dB+ carrier environments. The Navy will be constantly involved by downselecting various options and providing directors for shore and ship tests. Finally, we will investigate path planning and obstacle avoidance. There are many commercial uses for automated methods of recognizing commands and gestures in cluttered, challenging environments. Current products suffer from requirements for very simplistic conditions or expensive hardware. These requirements are too restrictive for a number of very promising commercial settings such as recognizing commands of small children. Carrier deck environments provide a challenging setting closely approximating realistic and challenging commercial settings, like a classroom full of small boys.