Projects

Detection of Small Space Debris in Low Earth Orbit

Agonic won first place in a NASA-sponsored contest that seeks to detect small space debris. Debris poses risk to crewed missions (such as in the ISS), scientific programs, and commercial interests. Agonic’s solution was judged to be feasible, effective, and cost-efficient. We are seeking partners for continued development. Announcement of winning concepts. Agonic submission was by Daniel Gebhardt.

Power Substation Load Prediction

We developed a system to predict load on a power grid substation 24 hours in advance. It uses weather forecast data, historical load data (considering an unknown amount of rooftop solar generation), and behavioral properties of the city’s population as input to its machine learning model. The service runs in the Microsoft Azure cloud and is available for customized adaptations for partners.

Precipitation Measurement Instruments

Existing rain and snow gauges fail in a variety of ways, leading to high maintenance costs and inaccurate watershed forecasting. Agonic developed two product prototypes that adopt modern embedded compute capabilities, sensor technology, and machine learning to fill this gap in environmental sensing, each with a novel approach. This effort includes design and fabrication of a prototype instrument (hardware, software, mechanical). Both product prototypes are in initial field evaluation with our partner.

Audio Signal Watermarking

The addition of a watermark to audio (or other signal) has many applications. Our project aims to minimize the audible changes created by the watermark by leveraging biologically-relevant ear models, driven by a machine learning process. It is in a research and development phase.

Asynchronous Neural Network Silicon Chip

Based on a patented concept, this collaborative project will engage academic and government institutions to produce a silicon chip architecture capable of running neural networks in a far more energy-efficient manner.