TECHNOLOGY PORTFOLIO
Innovative Solutions to Practical Problems
μSMET
μSMET is an autonomous, electrically-driven robotic mobility platform that can carry US Army light infantry personnel and payload quickly and quietly to and from conflict areas and rally points anywhere from narrow trails to urban alleys. Such a platform can be used in a number of last-mile applications in the battlefield. The robot also enables some unique force protection or offensive capabilities in the theater.
Heartbeat and Breathing Detection for Human Presence Indication
US Customs and Border Protection currently only scans about 15% of commercial cargo entering through land ports and 3% of the containers entering through sea ports with non-intrusive inspection equipment. Of particular concern is the prospect of terrorists entering the U.S. in an unscanned truck. M-Vision’s technology looks for human presence by heartbeat and breathing detection. A MEMS flex-reel sensor is taped onto sealed-containers and is periodically monitored for signals typical of human heartbeat and breathing. Our patented non-linear signal processing is used to overcome the issue of weak signal in the presence of strong background noise (road, wind, etc.).
Real-Time Measurement of Driver Distraction from Facial Video Streams
Distracted driving is dangerous, claiming 3,142 American lives in 2019. Distracted driving includes anything that takes your attention away from the road and the task of safe driving. Detecting distracted drivers is an important enabling technology in many advanced driver-assistance systems (ADAS) and connected-automated driving systems (CADS). M-Vision’s real-time measurement of driver distraction from facial video streams is a patented technology. Driver distraction is measured by an index called “time-off-road” — the amount of time in seconds that the driver is gazing at something else other than the road. Drivers are continuously monitored by a camera and their gaze is automatically estimated using a computer vision algorithm. This algorithm uses a non-linear subspace method to classify the driver’s pose into one of seven. The pose classification accuracy was shown to be up to 95% in independent testing by NHTSA and the Transportation Research Center (TRC).
Ultra-Wideband Multi-Spectral Target Detection System
In many Military, Automotive, and Homeland Security applications, the automatic detection of targets is an enabling technology. Some of the targets are deliberately concealed from detection. The Ultra-Wideband Multi-Spectral Target Detection System is successful due to its spectral adaptiveness — our system can image the same field-of-view anywhere from 400 nm to 20 μm. The signal-to-noise ratio of targets of interest are maximized by illuminating them in appropriate wavelengths, modulating this illumination with a sonic boom, and then shaping the target’s spectral characteristics in the image using polarizers and filters. A computer vision algorithm automatically detects targets of a certain shape listed in a database. The algorithm uses algebraic invariants in order to detect database targets irrespective of their orientation and scale. The system is embedded in a PC101 SBC and outputs results every 100 milliseconds.