Max Robotics — Global Service Robot Directory + US Certification

About

Rowan McAllister is a leading researcher at the intersection of safe reinforcement learning (RL), model-based control, and autonomous robotics. Their work tackles the fundamental challenge of deploying learning-based systems in the real world, where safety, data efficiency, and robustness are paramount. McAllister’s most influential contributions include the development of **SAVED** (Safety Augmented Value Estimation from Demonstrations), a model-based RL framework that enables robots to learn from sparse-cost tasks while maintaining safety constraints—a breakthrough that has garnered over 90 citations. They also pioneered **model-based meta-RL for aerial vehicles** (94 citations), enabling drones to autonomously adapt to the unpredictable dynamics of carrying suspended payloads, a critical capability for logistics and search-and-rescue. More recently, McAllister introduced **In-Distribution Barrier Functions**, a self-supervised method that filters out-of-distribution states to prevent catastrophic failures in learned controllers. Their work on **Risk-Aware Prediction (RAP)** further advances robust planning under uncertainty. With a career spanning hierarchical robot reconfiguration, Bayesian learning, and resilient navigation, McAllister’s research is defining how autonomous systems can operate safely and effectively in complex, unstructured environments—making them a pivotal figure in the future of trustworthy embodied AI.