[Paper Review] Towards an ASP-Based Architecture for Autonomous UAVs in Dynamic Environments (Extended Abstract)

Marcello Balduccini, William C. Regli+1 more|arXiv (Cornell University)|May 21, 2014

Logic, Reasoning, and Knowledge16 references3 citations

TL;DR

This paper proposes an ASP-based agent architecture for autonomous UAVs that integrates network-aware reasoning, centralized and decentralized planning, and real-time execution monitoring to handle dynamic, communication-constrained environments. By modeling network reliability and connectivity as part of the reasoning process, the system enables UAVs to autonomously re-plan and adapt to unexpected failures—reducing mission length by up to 30% and staleness by 50–100%.

ABSTRACT

Traditional AI reasoning techniques have been used successfully in many domains, including logistics, scheduling and game playing. This paper is part of a project aimed at investigating how such techniques can be extended to coordinate teams of unmanned aerial vehicles (UAVs) in dynamic environments. Specifically challenging are real-world environments where UAVs and other network-enabled devices must communicate to coordinate -- and communication actions are neither reliable nor free. Such network-centric environments are common in military, public safety and commercial applications, yet most research (even multi-agent planning) usually takes communications among distributed agents as a given. We address this challenge by developing an agent architecture and reasoning algorithms based on Answer Set Programming (ASP). Although ASP has been used successfully in a number of applications, to the best of our knowledge this is the first practical application of a complete ASP-based agent architecture. It is also the first practical application of ASP involving a combination of centralized reasoning, decentralized reasoning, execution monitoring, and reasoning about network communications.

Motivation & Objective

• Address the challenge of coordinating UAVs in dynamic, network-constrained environments where communication is unreliable and costly.
• Overcome the limitation of traditional AI planning that assumes reliable, free communication among agents.
• Develop a unified agent architecture that combines centralized mission planning, decentralized execution, and real-time network-aware reasoning.
• Enable autonomous adaptation to unexpected network failures through online explanation and re-planning.
• Integrate mission-level goals with network state reasoning to improve mission robustness and efficiency.

Proposed method

• Employ Answer Set Programming (ASP) as the core reasoning engine to represent complex knowledge, constraints, and dynamic behaviors.
• Model network connectivity and failures using logical rules that capture radio range, node failures, and communication reliability.
• Implement a hybrid architecture with centralized mission planning and decentralized local re-planning based on real-time observations.
• Use explanation facilities in ASP to infer the most plausible causes of communication breakdowns (e.g., relay node failure) from observed network states.
• Apply choice rules and optimization statements in ASP to generate optimal, goal-directed action sequences under constraints.
• Integrate the system with the CORE network emulator to simulate realistic, unreliable communication environments.

Experimental results

Research questions

• RQ1 How can UAVs autonomously re-plan when communication links fail unexpectedly in dynamic environments?
• RQ2 To what extent can network-aware reasoning improve mission efficiency and robustness in UAV coordination?
• RQ3 Can ASP-based reasoning effectively model the interplay between mission goals, UAV behavior, and network state dynamics?
• RQ4 How does integrating network modeling into the planning process reduce mission staleness and execution time?
• RQ5 What role does online explanation of failures play in enabling autonomous adaptation without pre-specified contingency behaviors?

Key findings

• The proposed ASP-based architecture enables UAVs to autonomously detect and explain communication failures, such as relay node outages, using logical inference.
• In experimental evaluations, the system reduced mission length by up to 30% compared to non-network-aware approaches.
• The approach achieved a reduction in total staleness between 50% and 100%, indicating significant improvements in mission timeliness and coordination.
• UAVs successfully re-planned and reconnected after unexpected relay failures, demonstrating robust adaptation without pre-defined contingency behaviors.
• The integration of network state reasoning into the planning process allowed for more resilient and efficient mission execution in dynamic, unreliable communication environments.
• The system demonstrated emergent, sophisticated behaviors such as mutual reconnection and coordinated path adjustment through online reasoning, not hard-coded rules.