The ATLAS experiment at the Large Hadron Collider employs resistive plate chambers (RPCs) for the level-1 muon trigger system in the barrel detector region. Excellence performance of the muon trigger is crucial for success of the ATLAS physics programme. New results detailing performance of the ATLAS RPC detector and trigger will be discussed. Next, plans for future upgrades of the RPC detector and trigger will be presented. These upgrades include the entirely new FPGA-based level-1 muon trigger system which will allow development of more powerful, more flexible trigger algorithms. This seminar will present our recent study of the FPGA-based neural network regression models. This study indicates that the RPC trigger algorithm using our neural network regression may significantly improve rejection of the dominant source of background events due to mismeasurement of muon momenta. FPGA simulation results of the neural network model will be presented. These results show that the FPGA resource usage and latency of this model are well within the requirements of the future level-1 trigger system. Other potential applications for this FPGA-based neural network model will also be briefly discussed.