Cross-platform Learning-based Fault Tolerant Surfacing Controller for Underwater Robots

2025-02-10

AI Summary

Sudden actuator failures in underwater robots often cause uncontrolled descent, necessitating robust fault-tolerant ascent control without relying on fault detection or isolation. Method: This paper proposes a learning-based, cross-platform fault-tolerant ascent controller trained via a PPO-based end-to-end reinforcement learning framework. It integrates dynamic models of heterogeneous underwater platforms (e.g., AUVs, U-CAT) and incorporates a partial policy transfer mechanism to enable knowledge reuse across platforms, coupled with Sim-to-Real transfer for physical deployment. Contribution/Results: To the best of our knowledge, this is the first approach achieving fault-tolerant ascent control without explicit fault identification. It significantly improves policy generalizability across heterogeneous platforms and training efficiency. Experimental evaluation demonstrates an 85.7% real-world ascent success rate—28.6 percentage points higher than the baseline—and validates effectiveness across three simulation environments and a physical U-CAT platform.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Multiagent Systems: Adversarial Agents
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

Abstract

In this paper, we propose a novel cross-platform fault-tolerant surfacing controller for underwater robots, based on reinforcement learning (RL). Unlike conventional approaches, which require explicit identification of malfunctioning actuators, our method allows the robot to surface using only the remaining operational actuators without needing to pinpoint the failures. The proposed controller learns a robust policy capable of handling diverse failure scenarios across different actuator configurations. Moreover, we introduce a transfer learning mechanism that shares a part of the control policy across various underwater robots with different actuators, thus improving learning efficiency and generalization across platforms. To validate our approach, we conduct simulations on three different types of underwater robots: a hovering-type AUV, a torpedo shaped AUV, and a turtle-shaped robot (U-CAT). Additionally, real-world experiments are performed, successfully transferring the learned policy from simulation to a physical U-CAT in a controlled environment. Our RL-based controller demonstrates superior performance in terms of stability and success rate compared to a baseline controller, achieving an 85.7 percent success rate in real-world tests compared to 57.1 percent with a baseline controller. This research provides a scalable and efficient solution for fault-tolerant control for diverse underwater platforms, with potential applications in real-world aquatic missions.

Problem

• Develops fault-tolerant surfacing controller for underwater robots.
• Uses reinforcement learning without identifying actuator failures.
• Enables cross-platform control policy transfer for efficiency.

Innovation

• Reinforcement learning control
• Transfer learning mechanism
• Cross-platform fault tolerance