A hierarchical framework for smooth and safe trajectory generation and finite-time tracking of greenhouse robots

To ensure the safety and reliability required for autonomous inspection in greenhouses—where high-precision tracking is essential, this paper proposes a two-layer hierarchical framework that couples smooth trajectory generation with safety, finite-time tracking for greenhouse mobile robots. The upper layer generates a smooth, dynamically feasible trajectory, while the lower layer enforces its execution in finite time under explicit safety and state constraints. In the trajectory generation layer, we formulate reference trajectory as minimum-jerk optimization constrained by safety corridors, guaranteeing strict geometric compliance with benches and greenhouse boundaries while avoiding abrupt changes in velocity and acceleration. The tracking layer implements a finite-time composite trajectory tracking controller equipped with explicit state constraints and a finite-time disturbance observer to compensate and estimate model uncertainties and external disturbance and preserving safety margins. Extensive simulations and real-world greenhouse deployments confirm that the planner consistently produces continuously differentiable paths, and the trajectory tracking controller attains high tracking accuracy while markedly reducing collision risk. The proposed hierarchical framework therefore delivers a practical, reliable navigation solution adapted to the structured and cluttered nature of greenhouse environments.