Utkarsh Grover (University of South Florida, Tampa, USA), Ravi Ranjan (Florida International University, Miami, USA), Mingyang Mao (University of South Florida, Tampa, USA), Trung Tien Dong (University of South Florida, Tampa, USA), Satvik Praveen (University of South Florida, Tampa, USA), Zhenqi Wu (University of South Florida, Tampa, USA), J. Morris Chang (University of South Florida, Tampa, USA), Tinoosh Mohsenin (Johns Hopkins University, Baltimore, USA), Yi Sheng (University of South Florida, Tampa, USA), Agoritsa Polyzou (Florida International University, Miami, USA), Eiman Kanjo (Nottingham Trent University, Nottingham, United Kingdom, Imperial College London, London, United Kingdom), Xiaomin Lin (University of South Florida, Tampa, USA)

Deploying foundation models in embodied edge systems is fundamentally a systems problem, not just a problem of model compression. Real-time control must operate within strict size, weight, and power constraints, where memory traffic, compute latency, timing variability, and safety margins interact directly. The Deployment Gauntlet organizes these constraints into eight coupled barriers that determine whether embodied foundation models can run reliably in practice. Across representative edge workloads, autoregressive Vision-Language-Action policies are constrained primarily by memory bandwidth, whereas diffusion-based controllers are limited more by compute latency and sustained execution cost. Reliable deployment therefore depends on system-level co-design across memory, scheduling, communication, and model architecture, including decompositions that separate fast control from slower semantic reasoning.

Sign up or login to leave a comment