The advances of technology have made sensor networks an attractive approach for detecting events present in a geographic region. Most of recent studies have focused on stationary sensor networks where sensor nodes are homogeneous with fixed locations. However, if the monitored region is significantly larger than one sensor's sensing area, it will need a large number of sensors to cover the whole region. In contrast, this paper proposes a novel sensing architecture for target detection using both mobile and stationary sensors in a sparse network. In the network, mobile and stationary sensors collaborate to detect the presence/absence of a target. Mobile sensors are responsible for both sampling the environment and collecting sensing data from stationary sensors. The collected data are fused to arrive at a consensus decision to determine whether the target is present. The paper proposes an effective approach for selecting routes with length as short as possible for mobile sensors while the constraints on the detection probability and false alarm rate are satisfied. The effectiveness of the proposed approach is shown by extensive simulations.