Observer-Based Robust Model Predictive Control for Safe Trajectory Tracking of Autonomous Vehicles Under Dynamic Constraints

Abstract

The safe deployment of autonomous vehicles depends on control architectures that can simultaneously manage uncertain state information, actuator limitations, environmental disturbances, and time-varying operational constraints. This paper presents a system-level examination of observer-based robust model predictive control for trajectory tracking under dynamic constraints. The discussion integrates estimation theory, constrained predictive control, and reachability-based safety reasoning to analyze how observer dynamics and robust constraint handling interact within a unified control framework. Rather than focusing on a single algorithmic derivation, the paper highlights architectural trade-offs among state reconstruction accuracy, computational complexity, actuator authority, and safety margin management. The analysis covers dynamic envelope constraints that adjust with road geometry, friction conditions, obstacle motion, and vehicle state, as well as the structural implications of embedding these constraints in real-time optimization. Deployment concerns such as embedded computation, communication latency, sensor fusion, and fail-operational behavior are examined in relation to functional safety standards. The paper further considers validation approaches, fairness across heterogeneous road users, and policy implications for certification and regulatory oversight. Through this interdisciplinary perspective, the paper argues that observer-based robust model predictive control should be understood not solely as a numerical control method but as a socio-technical system requiring coordinated design across estimation, optimization, infrastructure, governance, and safety assurance.