In search and rescue applications it is important that mobile ground robots can verify whether a potential target/victim is indeed a target of interest. This paper describes a novel approach to multi-robot target verification of multiple static objects. Suppose a team of multiple mobile ground robots are operating in a partially known environment with knowledge of possible target locations and obstacles. The ground robots' goal is to (a) collectively classify the targets (or build models of them) by identifying good viewpoints to sense the targets, while (b) coordinating their actions to execute the mission and always be safe by avoiding obstacles and each other. As opposed to a traditional next-best-view (nbv) algorithm that generates a single good view, we characterize the informativeness of all potential views. We propose a measure for the informativeness of a view that exploits the geometric structure of the pose manifold. This information is encoded in a cost map that guides a multi-robot motion planning algorithm towards views that are both reachable and informative. Finally, we account for differential constraints in the robots' motion that prevent unrealistic scenarios such as the robots stopping or turning instantaneously. A range of simulations indicates that our approach outperforms current approaches and demonstrates the advantages of predictive sensing and accounting for reachability constraints.