TUTORIAL SESSION

Multi-Objective Autonomous Robotics for Management and Operations of Complex Logistics Systems

Dr. Carlos R. B. Azevedo
Ericsson Research, Brazil
carlos.azevedo@ericsson.com

Duration

3 hours

Abstract

In this tutorial, we will explore the interplay between multi-objective computational intelligence approaches and autonomous robotics for managing complex logistics operations. In typical scenarios, each node of a supply network - e.g. suppliers, warehouses, retailers - has little information about the current states of the nodes affecting them. Such lack of information leaves no choice but to react to incoming orders from other nodes by relying on static business policies in order to fulfill production, inventory replenishment, storage, and delivery requests. As a consequence, achieving adequate service levels under acceptable costs can become very challenging. Vendor-Managed Inventory (VMI) approaches, where the supplier assumes full responsibility of replenishing the retailers inventories, can alleviate this problem by taking advantage of increased integration between nodes and conducting proactive automated operations for reducing supply-demand gaps. This tutorial will explain how to model and simulate an intelligent, automated VMI strategy allowing for efficient real-time integration of warehouse operations with multi-retailer inventory replenishment tasks. We will describe the simulated environment built with V-REP, a comprehensive robot simulator, where an autonomous industrial robot receives instructions generated by a multi-objective evolutionary algorithm running in real-time, aiming to simultaneously maximize profit and minimize shortage and surplus risks, while deciding on-the-fly which and how many products should be delivered to which retailers and when. Action selection is performed by monitoring inventory levels and stochastic customer ordering events at the retailers. In summary, the tutorial goals are (1) to expose multiple criteria decision-making concepts for supporting real-time expression of different degrees of risk aversion preferences; (2) to expose how retailer inventory control can lead to a sequence of warehouse pickup and delivery problems with deadlines; and (3) to demonstrate the cascading effects on the simulated robot behavior of demand events in the retailers side that can disrupt the supplier’s service level capacity.

Objectives

1. Expose multiple criteria decision-making concepts for handling complex scenarios in operations management;
2. Expose modeling, automation, and integration concepts of warehouse management with retailer inventory control;
3. Demonstrate the effects of the approach on integrated warehouse operation using a virtual robotics simulator.

Timeline

- Introduction to multi-objective optimization and multiple criteria decision-making (15 min);
- Introduction to warehouse management modeled as pickup and delivery with time windows problems (45 min);
- Introduction to mobile and service robotics (30 min);
- Present the V-REP robotics 3D simulator (15 min);
- Discuss the modeling and implementation aspects of the proposed solution (30 min);
- Present results and demonstrations (45 min).