Deceptive Decision-Making Under Uncertainty

TL;DR

This paper introduces a method for autonomous agents to deceive observers about their true goals in complex environments, using Markov decision processes and maximum entropy modeling to generate tunable deceptive strategies.

Contribution

It presents a novel approach combining maximum entropy modeling and linear programming to create efficient, probabilistically constrained deceptive behaviors in autonomous agents.

Findings

Agents can exhibit diverse deceptive behaviors.

The approach outperforms baseline methods in user studies.

Real-world case study demonstrates practical applicability.

Abstract

We study the design of autonomous agents that are capable of deceiving outside observers about their intentions while carrying out tasks in stochastic, complex environments. By modeling the agent's behavior as a Markov decision process, we consider a setting where the agent aims to reach one of multiple potential goals while deceiving outside observers about its true goal. We propose a novel approach to model observer predictions based on the principle of maximum entropy and to efficiently generate deceptive strategies via linear programming. The proposed approach enables the agent to exhibit a variety of tunable deceptive behaviors while ensuring the satisfaction of probabilistic constraints on the behavior. We evaluate the performance of the proposed approach via comparative user studies and present a case study on the streets of Manhattan, New York, using real travel time distributions.