Abstract: With the rapid penetration of distributed renewable generation, energy storage and heterogeneous loads, the scheduling of distributed energy systems (DESs) exhibits complex operational characteristics featured by deep coupling of multi-energy flows, multi-timescale dynamic coordination, and multi-agent interactions. Multi-source uncertainties propagate and disturb across energy carriers and time horizons, reducing system safety margins and increasing the difficulty of maintaining feasible system operation. As a result, DES scheduling is driven to evolve from deterministic cost-oriented optimization toward risk-aware and feasibility-guaranteed robust decision-making. Firstly, this paper summarizes the scheduling modeling framework for DESs. Combined with the propagation characteristics of uncertain disturbances in multi-energy conversion chains and network constraints, the main sources of uncertainties as well as their cross-carrier and cross-time-horizon propagation features are systematically analyzed. Secondly, modeling paradigms including scenario sets, uncertainty sets and distributional ambiguity sets are summarized, and their different impacts on scheduling feasibility and risk control are compared. Special emphasis is placed on the model formulation, solution methodologies and applicable scenarios of robust optimization, multi-stage adaptive robust optimization and distributionally robust optimization for DES scheduling. Key issues including non-anticipativity, full-scenario feasibility, conservativeness control and out-of-sample risk management are elaborated, together with their application boundaries concerning safety, economy and data dependence in DESs. Finally, future research directions for robust scheduling of DESs under uncertainties are discussed.