Max Robotics — Global Service Robot Directory + US Certification

About

Mohammad Mahdavian is a researcher specializing in human motion prediction and human-robot interaction, with a particular focus on developing advanced deep learning architectures to anticipate and respond to human movement in real time. His most significant contribution, the STPOTR framework, introduced a non-autoregressive transformer architecture capable of simultaneously predicting human trajectory and body pose — a dual capability that substantially advances robots' ability to perform follow-ahead tasks. This work, accumulating 25 citations since its 2023 publication, demonstrates both technical innovation and practical impact in autonomous robotic systems. Mahdavian has also made notable strides in addressing the diversity problem inherent in human motion forecasting. His DMMGAN model leverages attention-based generative adversarial networks to produce multiple plausible future motion sequences for 3D human joints, moving beyond the single-prediction limitations of earlier approaches. With 13 citations, this work highlights his commitment to capturing the inherently stochastic nature of human behavior. Across his research portfolio, Mahdavian consistently pushes the boundaries of what robots can anticipate about human movement, making meaningful contributions to safer, more responsive human-robot collaboration systems.