Sensor networks are being used to implement different types of sophisticated emerging applications, such as those aimed at supporting ambient intelligence and surveillance systems. This usage is enhanced by employing sensors with different characteristics in terms of sensing, computing and mobility capabilities, working cooperatively in the network. However, the design and deployment of these heterogeneous systems present several issues that have to be handled in order to meet the user expectations. The main problems are related to the nodes' interoperability and the overall resource allocation, both inter and intra nodes. The first problem requires a common platform that abstracts the nodes' heterogeneity and provides a smooth communication, while the second is handled by cooperation mechanisms supported by the platform. Moreover, as the nodes are supposed to be heterogeneous, a customizable platform is required to support both resource rich and poorer nodes. This paper analyses surveillance systems based on a heterogeneous sensor network, which is composed by lowend ground sensor nodes and autonomous aerial robots, i.e. Unmanned Aerial Vehicles (UAVs), carrying different kinds of sensors. The approach proposed in this work tackles the two above mentioned problems by using a customizable hardware platform and a middleware to support interoperability. Experimental results are also provided.