In recent years, the improvement of quality of life in cities through the reduction of localpollutant emissions and traffic has been an increasingly important topic for city planners andresearchers. This leads to classical practices such as using vans for deliveries to be restricted andother modes preferred in some areas such as campuses, pedestrian zones or inner-city districts.Recent technological developments and experiments point to the potential of automated lastmile deliveries using small battery-powered robots. We therefore investigate a two-echelonhybrid van-based robot delivery system for last-mile logistics operations.In existing research, using robots for second-echelon delivery has been considered. For example,a two-echelon delivery model (2EVRP) is proposed in [1]. At the first echelon, vans transportpackages from a large depot to small local robot hubs. At the second echelon, robots deliveritems to customers. A van-based robot delivery model (VRD) concept with time windows isstudied in [2]. Here, robots are transported in the vans to perform second-level route deliveries.According to a sensitivity analysis performed in [3], each of these delivery models have theiradvantages. Depending on fixed costs of vans, robots or satellites, as well as other factors, oneof the models may perform better than the other.We propose a hybrid delivery model to select the optimal distribution strategy consideringconventional vans, inner city hubs with robots, and delivery vans carrying robots, potentiallyimproving distribution efficiency. We model the problem as mixed-integer program and proposean adaptive large neighborhood search (ALNS) to solve large instances efficiently. We conductcomputational experiments to evaluate how our proposed model performs and how differentcharacteristics (size, time) of the access restrictions impact its performance. We also evaluatehow different characteristics impact solution structure