The upcoming sPHENIX experiment, scheduled to start data taking at the BNL Relativistic Heavy Ion Collider in 2023, and the future EIC experiments will employ sophisticated state-of-the-art, high rate detectors to study high energy heavy ion and electron-ion collisions, respectively. The resulting large volumes of raw data far exceed available DAQ and data storage capacity. To meet this challenge, we propose to develop a selective streaming readout system comprising state-of-the-art AI-based fast data processing and autonomous detector control systems. This will allow to effectively sample the full high energy collision events delivered by the accelerators while maintaining the final data throughput for offline storage at a manageable level within the available DAQ bandwidth, storage, and computing capacity. This project designs real-time AI-based algorithms that operate on high-rate data streams and allow the identification of important rare physics events from abundant backgrounds in the sPHENIX's p+p and p+Au collisions, as well as in the future EIC experiments, such as the one proposed by the ECCE consortium. We will co-design physics-aware high-speed deep neural networks that automatically perform complex tasks of collision event reconstruction and analysis, monitor and calibrate the beam interaction points, and align detectors in real-time. Demonstrating such a full system integration will be the first step in autonomous control loops of powerful online AI algorithms for large-scale, complex high-energy nuclear physics experiments.
