VISUS LLC — National Science Foundation SBIR Phase II: I

Description

This Small Business Innovation Research (SBIR) Phase II project will accelerate the adoption of data intensive precision agriculture, increasing yields while decreasing farm inputs such as fertilizers and pesticides. This project removes the software bottleneck (time and labor) in processing large aerial surveys taken by Unmanned Aerial Systems, enabling a cost-effective and timely process to deliver actionable information to farmers. Using frequent high-quality aerial scans, farmers may optimize the use of fertilizers and more finely control the amount of pesticides and herbicides necessary to increase crop yield. Furthermore, farmers mitigate costs and losses by being able to spot problem areas, minimize the spread of plant diseases, and identify issues such as standing water, irrigation malfunctions, and persistent automated machinery errors in planting or cultivation. This project provides special benefit for rural customers having inadequate internet infrastructure by eliminating the need to upload massive imagery to the cloud for processing. The technology is part of a broad initiative in agriculture addressing the need for large increases in food production by 2050 in response to the projected growth of the world?s population to over 9 Billion people. This project will continue development of algorithms for on-the-fly orthorectification, stitching, and normalization of aerial image mosaics and their deployment in an easy-to-use software prototype. The Phase I already demonstrated industry-leading speeds for such image processing. The technology behind this research project is designed from the ground up to process massive data with less memory and increased speed relative to other approaches, enabled by a proprietary streaming image representation, that allows multichannel gigapixel and terapixel images to be treated as ordinary images. This Phase II supports new extensions to the software that simplify and accelerate delivering a stitched and analyzed map, such as prioritizing computation in regions of the image that a customer is exploring. This would effectively eliminate the delay between image acquisition on unmanned aerial vehicles and when it can be used. Crop consultants have identified this as a transformative capability, as it enables ground-truthing information derived from aerial imagery in the same field visit, saving time and labor. The performance gains in compute-limited environments supported by this project are a key link between new capabilities to gather information and a farmer?s ability to utilize it to increase productivity while reducing costs.