There is an increasing research interest in solving routing problems in Wireless Sensor Networks (WSN) and most of the existing routing protocols for WSN consider homogeneous nodes wherein, all sensor nodes have the same capabilities in terms of sensing, communication and computation capabilities. However, deployment of a homogeneous sensor network may suffer from poor performance. In this paper, we present a novel QoS routing scheme that adaptively learns an optimal routing strategy, depending on multiple optimization goals. We base our approach on Ant Colony Optimization (ACO) technique which is insensitive to initial setting, thus ideal in the context of WHSNs. The key feature of the protocol is its ability to meet diverse QoS requirements posed by different kinds of traffic generated due to heterogeneous nature of nodes thus maximizing network performance and utilization. Experimental results suggest a significant gain in the performance of proposed protocol over AODV protocol.