tinyGIANT straddles the worlds of technology and social impact - this blog post, part of a series, explores our work in the technology sector.
Tyler has served as the VP of Product for two Bay Area start-up tech companies focused on AI/ML and robotics: OSARO and Airspace Systems. The products and systems described below (organized by subject matter) were developed either under his product management leadership while at these companies or during consulting engagements through tinyGIANT on other technology-focused products.
AI/ML Software
Airspace Systems’ machine vision algorithms detect airborne drones, track their position through the sky, and continually evaluate and validate detected drones in order to reject false positives. The machine vision system runs locally, both onboard the in-flight Interceptor (optimized for a lightweight compute platform) and within the terrestrial detection and tracking “Pods”.
OSARO’s machine learning-based vision and control algorithms differentiate individual items in jumbled scenes, determine which item to grasp next and the orientation in which to grasp it, provide in-flight dimension and weight information regarding the object “in-hand”, and direct the precise and optimized placement of the item at its destination.
Here is Tyler, discussing how OSARO's machine vision system operates:
Robotics
OSARO’s pick-and-place robots are deployed to warehouse, logistics, and e-commerce facilities, where they perform dull, dirty, and dangerous operations that are frequently difficult to staff with human operators. The robot control software, tightly coupled with the vision stack and integrated suite of sensors, direct the operation and behavior of the industrial robots as they adapt, in real time, to the vast range of items they are expected to handle.
Here is a video of OSARO’s automated bagging system in operation:
And another video of Tyler discussing how another of OSARO’s products, a sorter induction system, operates:
Autonomous Vehicles
Airspace System’s Interceptor is a fully autonomous unmanned aerial vehicle (UAS), aka “drone”, responsible for protecting sensitive airspace (think: airports, stadiums, critical infrastructure) from unauthorized drones. Integrating telemetry from terrestrial sensors and onboard machine vision system, the Interceptor locates intruding drones and safely removes them from the sky.
Here is a video of an early prototype of the Interceptor in action:
Cloud & Edge & IoT
Airspace Galaxy (UAS detection and tracking system) relies on a constellation of geographically distributed “Pods”, each of which contains radio frequency (RF) sensors, cameras, and local compute. Airborne drones are detected using RF sensors and pinpointed with cameras running Airspace’s machine vision algorithms on local hardware. The telemetry and video data from all these disparate feeds are correlated, streamed in real time, stored locally, and uploaded to the cloud for historical analysis and forensics.
Here is a video of Airspace Galaxy in action during San Francisco’s Fleet Week:
Web & Mobile Applications
UNICEF’s Emergency Preparedness Platform is a web portal through which all 190 of UNICEF’s country offices prepare for disasters and plan their coordinated response – user research, conceptual design, and system prototyping by tinyGIANT (project link).
Airspace Galaxy (UAS detection and tracking system), comprising a web and mobile app, provides notifications, identification information, a 3D map-based visualization, and video streaming interface for drones detected in sensitive airspace - for use by security personnel and first responders.
Here is a video of Airspace Galaxy’s web-based visualizations:
API
OSARO’s Pick-and-Place API defines the commands that must be implemented by a customer's warehouse control system (WCS) in order to interoperate with an OSARO robot performing pick-and-place tasks at a station within a larger automated warehouse system.
Hardware
Airspace’s Interceptor and Hangar comprise proprietary hardware developed, prototyped, and fabricated internally: airframe, embedded systems, integrated sensors and compute, net-capture system, enclosure, launch system. Below are CAD renderings of each system:
Panasonic’s UV Water Sterilizer prototype developed by tinyGIANT consisted of a 3D printed housing, proprietary embedded systems, and sterilizing UV-C light bulb (project link). Here is a photo of the fully-functional system prior to user testing in India:
Patents
Airspace System’s net-capture system is described in this patent (US USD865116S1). Here is a photo of it in action:
Airspace System developed this foundational patent (US US20190103030A1) to describe a system for identifying the remote identity of airborne drones (and operators) prior to Tyler sitting on the Federal Aviation Administration’s (FAA) UAS Identification and Tracking Advisory and Rulemaking Committee dedicated to the same subject (the final report submitted by the committee to the FAA can be found here – link).
Comments