RI: Small: Collaborative Research: Cooperative Autonomous Vehicle Routing under Resource and Localization Constraints
This project aims to develop novel algorithms required to deploy Unmanned Vehicle (UV) networks with resource constraints in Global Positioning System (GPS) denied environments. The methods developed in this project will be useful in a wide variety of applications of national importance such as disaster management, border surveillance, monitoring of civilian infrastructure including oil pipelines, power grids, harbors, inland waterways, and intelligent transportation systems where GPS signals can be easily jammed either intentionally or unintentionally. The proposed research spans several areas including control, estimation, sensing, robotics and optimization. This project provides a rich opportunity for involving undergraduate and graduate students in the development of vehicle platforms, sensor networks, and in the implementation of the control and optimization algorithms. This project engages minority students in small research projects to motivate their interest in engineering and science. Enabling autonomous unmanned vehicles with a capability of navigating in GPS denied environments can aid in effectively monitoring large infrastructure systems, protect their structural integrity and functional reliability as well as provide ecological, societal and economic benefits, including better preservation of natural resources, reduced property damage and reduced loss of life.
This proposal addresses the following fundamental problem that arises while deploying unmanned vehicles in GPS-denied environments: Given a set of vehicles and targets to visit, find a path for each vehicle such that each target is visited at least once by some vehicle, the error in the position estimate of each vehicle at any time instant is within a given bound and an objective which depends on the travel and sensing costs is minimized. The specific technical objectives of this project are to: determine the minimal set of requirements that would render the system of vehicles observable over a time period, develop novel approximation and exact algorithms using cutting plane, rounding and Lagrangian dual methods for the optimization problems, and experimentally corroborate the performance of the proposed algorithms using large scale and hardware-in-the-loop simulations, and field demonstrations. It is anticipated that this project will significantly advance the state of art in the area of observability analysis for a team of cooperatively localizing vehicles, and in the area of tractable, approximation and exact algorithms for vehicle placement and path planning problems with resource and localization constraints. Novel cutting plane, rounding, and Lagrangian dual methods are expected to provide new insights into efficient ways of decomposing the difficulties in the vehicle placement and path planning problems, and will lead to good feasible solutions with approximation bounds. The proposed large scale simulation and experimental results will provide a new understanding of the influence of the different parameters (number of landmarks/vehicles/targets, bounds on acceptable position errors, onboard sensor type, different operational environments, and the speed of each vehicle) on the performance of the vehicle localization/path planning system.