Invited Speakers

Subset Approximation of the Pareto Frontier in Multi-Objective Search: Basic Notions and Perspectives

Jorge Baier

Pontificia Universidad Católica de Chile 
Abstract: Multi-objective search (MOS), which requires finding a Pareto frontier of paths on a search graph that minimizes a number of objective functions, is an important search problem with a number of real-world applications. This talk will present a recent compilation-based approach to finding subset approximations of the Pareto frontier of solutions. The approach takes a MOS problem P and produces another MOS problem P' whose Pareto frontier is a subset of that of P'. Solving P' can usually be solved faster than P. Also, they may provide 'diverse' solutions. The talk will give the intuitions of why these approaches may or may not work efficiently and why they may or may not produce diverse solutions.
Bio: Jorge Baier is an Associate Professor in the Department of Computer Science, and the Associate Dean of Engineering Education, at the School of Engineering of Pontificia Universidad Católica de Chile. His research interests are in the area of heuristic search, automated planning, and in the application of AI to improve the student's experience in higher education.

The Magic From Click To Delivery At Amazon: It's Search!

Scott Kiesel

Amazon Robotics
Abstract: Amazon tackles the problem of coordinating a heterogeneous fleet of hundreds of thousands of robots to go from a website click to a delivery at your door. While formulating the underlying optimization problem is possible (albeit impractical), instantiating and solving it at such scales is infeasible. At Amazon, we leverage problem decomposition and rely on proven methods to solve these smaller problems, and finally apply system engineering to reconstruct an executable solution to the original problem, all in real-time. Some of the most useful techniques in our tool belt are search-based methods. In this talk, we’ll discuss how we tackle decomposition, examine examples of where search is currently deployed, and where we see further opportunities for search to be used.
Bio: Scott Kiesel is a Senior Applied Scientist at Amazon Robotics. He earned his PhD from the University of New Hampshire (UNH) in 2016. His current research focus is large-scale multi-agent planning and execution algorithms and systems. At Amazon, he shepherds novel research projects from ideation through production deployment, working closely with Software Development Engineers.