Landmarks for Deception

Primary supervisor

Mor Vered

Co-supervisors

  • Ramon Fraga Pereira

 

In automated planning, landmarks are properties (or actions) that every plan must satisfy (or execute) at some point in every plan execution to achieve a goal. Landmarks are often used to build planning algorithms. This project aims to leverage the concept of landmarks and existing landmark based heuristics in order to create plans that may deceive potential observers. 

 

Student cohort

Single Semester
Double Semester

Aim/outline

The aim of the project is to develop and implement a novel, state-of-the-art deception planner and to evaluate its effect over classical planning benchmarks in multiple domains which include but is not limited to path planning. 

URLs/references

Pereira, Ramon Fraga, and Felipe Meneguzzi. "Landmark-based plan recognition." arXiv preprint arXiv:1604.01277 (2016).

Hoffmann, Jörg, Julie Porteous, and Laura Sebastia. "Ordered landmarks in planning." Journal of Artificial Intelligence Research 22 (2004): 215-278.

Masters, Peta, and Sebastian Sardina. "Deceptive Path-Planning." IJCAI. 2017.

Required knowledge

This project will require extensive programming with python as well as some knowledge of unix like operating system commands.