Formation control analyses the convergence of a group of mobile agents to predefined geometric patterns. In traditional approaches, it is assumed that each agent knows the exact position of certain members of the group with respect to a reference frame and the associated control laws are designed according to inter-robot relative positions. Designing a more decentralized scheme, this paper proposes a formation scheme, using Lyapunov techniques, considering that the local controllers of the agents can be equipped with distance and orientation sensors. The main result of the paper applies to certain distance-based potential functions with inter-robot collision avoidance and an arbitrary undirected formation graph. Also, the control law includes an integral-type control that eliminates the effects of the dead-zone of actuators in order to avoid the standard techniques of normalization. The control approach is analyzed for omnidirectional robots with numerical simulations and extended for unicycle-type robots with real-time experiments. Graphical Abstract

This paper presents the extension of leader‐follower behaviours, for the case of a combined set of kinematic models of omnidirectional and differential‐drive wheeled mobile robots. The control strategies are based on the decentralized measurements of distance and heading angles. Combining the kinematic models, the control strategies produce the standard and new mechanical behaviours related to rigid body or n‐trailer approaches. The analysis is given in pairs of robots and extended to the case of multiple robots with a directed tree‐shaped communication topology. Combining these behaviours, it is possible to make platoons of robots, as obtained from cluster space or virtual structure approaches, but now defined by local measurements and communication of robots. Numerical simulations and real‐time experiments show the performance of the approach and the possibilities to be applied in multirobot tasks.

This paper addresses the problem of controlling reentrant semiconductor fabrication lines. To focus the presentation, a virtual five-machine six-step mini-fab is used. This mini-fab developed by Intel in collaboration with ASU is intended to contain all of the features which make a real line difficult to control. Industry-like policies are discussed and compared with control-theoretic hierarchical policies.