DARPA project looks at using artificial intelligence to process massive amounts of data and images on the battlefield where it is collected
The Defense Advanced Research Projects Agency is searching for research projects that use artificial intelligence to process data where it is collected, the agency announced.
This kind of dispersed analysis of data is called “extraction at the edge.” It is necessary, because the amount of data and pixels modern sensors collect require massive amounts of energy and computing power to process.
This project hopes to use AI to process it on the edge.
This Artificial Intelligence Exploration Opportunity wants submissions of innovative basic or applied research concepts related to low power data extraction at the edge.
In the mid-80s, models for visual attention (object tracking) began to adopt biological inspiration to improve accuracy and functionality. Since approximately 2015, trends have shifted towards implementing models in deep Neural Networks (NNs), such as Resnet and VGGNet, because they greatly outperform conventional methods in both accuracy and generalizability.
To achieve high accuracy in such NNs today, AI processing for video is largely performed in the data center by spatial algorithms.
AI processing of video presents a uniquely difficult problem because the high resolution, high dynamic range, and high frame rates desired generate significantly more data than other edge sensing modalities.
To accommodate this large data stream, current deep neural networks (DNN), require hundreds of millions of parameters for tens of billions of operations to produce a single accurate AI inference.
Today’s current solution is incompatible with embedded implementation at the sensor edge due to power and latency constraints and the result is that embedded solutions for vision sensing at the mobile edge abandon state of the art accuracy in favor of even marginally accurate solutions that can operate within the size and power available.
To learn more about the project and to submit a proposal, visit the DARPA posting HERE.
