Velocity observer design for the consensus in delayed robot networks

The consensus problem for networks of multiple agents consists in reaching an agreement between certain coordinates of interest using a distributed controller. It may be desirable that all the agents find a consensus at a given desired leader coordinate (Leader Follower Consensus Problem LFCP), or it may be only necessary that they agree at a certain coordinates value (Leaderless Consensus Problem LCP). Consensus has many practical applications in robot networks systems, where the interconnection of the agents may present variable time delays, hence rendering the stability analysis and control design more complex. Another problem that may arise is the possible lack of velocity measurements. In this work, a Proportional plus damping injection (P + d) controller together with a linear velocity observer is introduced. Our approach is able to solve both the LFCP and the LCP in networks of robots modeled as undirected weighted graphs with unknown asymmetric (bounded) variable time delays. Local (semi global) asymptotic stability is proven and simulation results are provided to test the performance of the proposed scheme.