A Dynamic Leader-Follower Strategy for Multi-robot Systems

Since there is not a very suitable dynamical leader selection model for formations control of the multi-robot system, we propose an affection-based dynamic leader selection strategy. A leader selection modular is used to switch the leader autonomously according to the cognized environmental statement based on two virtual affections, disappointment and abashment. The disappointment of followers will increase with the time spending in the awful scenario. When it exceeds a certain threshold, the followers will broadcast an unsatisfied signal. The abashment value of leader changes with the unsatisfied signals received from the followers and flutters with its own status at the same time. When the abashment value goes beyond the threshold, the leader re-selection process will be triggered. The abashment threshold is determined by the theoretic calculation. Plenty of simulation results show that the multi-robot system with dynamic leader-follower strategy has more probability of getting out of the woods, and also promote the survivability rate in the leader falling situation by re-selecting a leader autonomously.