Receding horizon temporal logic control in dynamic environments

Abstract

We present a receding horizon method for controlling an autonomous vehicle that must satisfy a rich mission specification over service requests occurring at the regions of a partitioned environment. The overall mission specification consists of a temporal logic statement over a set of static, a priori known requests, a regular expression over a set of dynamic requests that can be sensed only locally, and a servicing priority order over these dynamic requests. Our approach is based on two main steps. First, we construct an abstraction for the motion of the vehicle in the environment by using input–output linearization and assignment of vector fields to the regions in the partition. Second, a receding horizon controller computes local plans within the sensing range of the vehicle such that both local and global mission specifications are satisfied. We implement and evaluate our method through experiments and simulations consisting of a quadrotor performing a persistent surveillance task over a planar grid environment.