Speakers
Description
Particle detectors rely on triggering logic to record meaningful events. Historically, most electronic triggers use field programmable gate arrays (FPGAs) due to their flexibility and relatively low cost compared to custom chips (ASICs). Nowadays, FPGAs have grown to hundreds of thousands of logic units, enabling them to act as neural networks, the caveat being that models need to be simplified to fit the combinatorial logic in what is called a field programmable neural array (FPNA). Implementing FPNAs as triggers offers potential advantages over traditional logic triggers, primarily the ability to classify complex patterns. This research project, internally known as CAIPORA (Compact Artificial Intelligence Placed on Reprogrammable Array), initially aimed at triggering supernova events and has found its first use not in astrophysics experiments but on car headlights. Approaching a local automotive headlight industry, we proposed using the CAIPORA technology to help them develop adaptive headlights, those that can turn off selected areas of the light beam so as not to obfuscate incoming drivers. We present the first year's results of this three-year FAPEMIG-funded partnership and discuss the advantages of this model of funding for both science and industry.
| I read the instructions above | Yes |
|---|