Receding Horizon Multi-Agent Deceptive Path Planner

Deceptive path planning enables autonomous agents to obscure their true goals from observers by deviating from an expected optimal path. Prior work largely solves full-horizon, end-to-end optimization for single agents, so planning complexity grows with the path length and size of the environment, and adaptation en route requires a new optimization. We propose a unified framework for deceptive path planning onboard the agent, computing over short-horizon candidate paths within a receding-horizon loop. The method processes an agent's position sensor information and develops a policy for the next trajectory plan. By parameterizing a user-defined cost that captures optimal planning and deception (and optionally includes constraints, trajectory smoothness, and coupling terms between agents), a Boltzmann framework yields stochastic policies that balance the tradeoff between optimal paths and deceptive deviation. Policies are updated locally and do not require learning or training. The level of deception and adherence to constraints can be dynamically tuned, enabling online adaptation to changes in goals and constraints. This step-by-step tuning opens the door to new forms of dynamic deception. In the multi-agent case, we develop a joint planner that is also tunable and can be applied as desired among all or subsets of agents. Single- and multi-agent simulation studies demonstrate the flexibility of our approach, maintaining deception while adapting as desired, avoiding the recomputation required by full-horizon methods, and supporting intuitive tuning via a small set of parameters.