A decentralized coordination strategy is developed to generate parallel motion of multiple mobile robots. All robots communicate with each other to form a robotic network the topology of which is modelled by a weighted digraph. The linear consensus algorithm is extended to design the coordinated control strategy, in which bounded velocity specifications are considered to avoid actuator saturation. A design method of information flow topology for two modes of parallel motion, leader-follower and leaderless, is investigated. The effectiveness of control laws and effects of different information flow topologies and sampling rates on the group behavior are shown in nontrivial computer simulations.