Optimal Mission Planning under Logical Constraints with Safety Gaurantees: An Application for Robotic CPSs
Authors :
Zhou, Yuchen
Maity, Dipankar
Conference : International Symposium on Networked Cyber-Physical Systems (NetCPS)
Date: September 19 - September 20, 2016
In this work, we address logical constraints satisfiability in cyber-physical systems (CPS). Even in a simplistic case, where more than one physical components are present, the system inherently possesses some underlying network structure and hence can be studied under a general networked-CPS (Net-CPS) framework. The underlying networked structure and the (a) causal inter-dependency among the components lead to some logical and temporal constraints that need to be satisfied over such Net-CPSs. We focus on a controller synthesis problem under such temporal-logic constraints; while the temporal logic represents the logical constraints, the controller synthesis address the problem of driving the physical components in an orderly manner to satisfy those constraints. Among the vast majority of ubiquitous Net-CPSs, we restrict our attention to robotics and autonomous car scenarios in order to motivate and validate our research and results. We present a novel approach to address these critical problems which combines metric temporal logic, timed automata and reachability analysis. We study the problem of human-robotic (or multi-robotic) motion planning under temporal logic conditions. Given a task, the team logically decomposes (through a suitably designed algorithm) the task into subtasks and each member of the team is assigned subtasks, which they perform according to their ability. We precisely address issues arising from nite time speci fications in such situations since the time constraints could be implicit. Consequently, the controller synthesis is much more complicated than without the time-boundedness specications. We also develop a reachability based approach in-order to guarantee real-time and nite-time safety in the presence of initial and unexpected uncertainties.
Download Full Paper