Research on data routing strategies for wireless sensor networks (WSNs) has largely focused on energy efficiency. However rapid advances in WSNs require routing protocols which can accommodate new types of energy source and data of requiring short end-to-end delay. In this paper, we describe a duty-cycle-based low-latency geographic routing for asynchronous energy-harvesting WSNs. It uses an algorithm (D-APOLLO) that periodically and locally determines the topological knowledge range and duty-cycle of each node, based on an estimated energy budget for each period which includes the currently available energy, the predicted energy consumption, and the energy expected from the harvesting device. This facilitates a lowlatency routing scheme which considers both geographic and duty-cycle information about the neighbors of a node, so that data can be routed efficiently and delivered to the sink as quickly as possible. Simulation results confirm that our routing scheme can deliver data to th...