Optimized coordination and scheduling of traffic evolving on complex guidepath networks

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

In many contemporary industrial and distribution applications, the material handling operations are executed automatically by fleet of robotized agents (e.g., mobile robots, vehicles circulating on elevated monorail systems, or complex arrays of gantry cranes) circulating over a guidepath network that is defined either by the physical structure of the material handling system (as in the case of the monorail and the gantry crane systems), or defined externally in an effort to separate the corresponding traffic from the surrounding environment. As these material handling systems increase in scale and operational complexity, there is an emergent need for a more systematic methodology that will support expedient and safe traffic for the traveling agents, while taking into consideration all the constraints and limitations that arise from the constricted nature of the underlying guidepath network and of the limited maneuvering capabilities of the agents themselves. Furthermore, similar traffic dynamics and control needs arise in the operational context of quantum computing systems. In that case, the ionized atoms (or qubits) that carry the elementary information to be processed in those systems, must be transported to different locations in order to interact with each other or with certain external fields, and this transport must be performed through a network of carefully defined and allocated tunnels (more formally known as ion traps); these tunnels and the corresponding allocation of them to the traveling qubits intends to protect the processed information by isolating the qubits from their surrounding environment and preventing unintended interaction among them. The successful completion of the proposed research program will develop a complete methodology for effective and efficient real-time coordination and scheduling of the traffic that takes place in the aforementioned application environments. In particular, the expected results will enable safe, orderly and expedient traveling for the system agents in the face of all the complexity of the underlying operations, which will further lead to higher levels of flexibility, robustness and productivity for these operations. Furthermore, the project will employ a graduate student who will carry out a Ph.D. thesis on the proposed set of problems, and the obtained results will be implemented on a high-fidelity simulator of a quantum computer that is developed by a group of researchers at the Georgia Tech Research Institute and have an active collaboration with the PI and the aforementioned graduate student. In addition, the PI will leverage his affiliation with the College-Industry Council on Material Handling Education in order to promote the obtained results to the relevant material handling industries. To effectively meet the research objectives outlined in the previous paragraph, the proposed research program will substantially complement and extend the current theory on the real-time coordination and scheduling of complex traffic systems. In the proposed modeling and analysis paradigm, the considered traffic management problems will be represented as sequential resource allocation problems where the running processes are the trips executed by the system agents, and the allocated resources are the various segments of the guidepath network that are needed for the execution of these trips. This allocation must be controlled for logical correctness, that in the considered problem context translates to the establishment of safety and liveness for the underlying traffic, and also for expediency, which translates to delay minimization and high throughput for the requested transports. All these requirements will be met through the employment, extension and integration of ideas and results coming from the areas of (i) logical control of complex resource allocation systems, (ii) (combinatorial) scheduling theory, and (iii) the embedding of these results in a novel model predictive control framework.