USV Swarm Formation Control With CEDRL Collision Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in effectively controlling large-scale unmanned surface vessel (USV) swarms, particularly in quickly changing formation patterns and avoiding collisions, with limited autonomous and collaborative capabilities and complex motion controllers.
Innovation Solution
A method and system utilizing Collaborative Exploration Deep Reinforcement Learning (CEDRL) to design a desired formation pattern based on a hierarchical virtual leader strategy, update USV locations via a local consensus strategy, and employ a surface vessel control decision-making network to navigate USVs to their desired locations, incorporating a reward function for collision avoidance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional swarm formation control algorithms are used, then basic coordination is achieved, but autonomous control capability is limited and flexibility is poor
Solution Approach 1:
Each USV is equipped with an autonomous decision-making network that independently processes local sensor data and generates control commands without requiring complex centralized coordination. The USVs self-organize into formations through local interactions, achieving adaptive formation patterns while keeping individual controller complexity manageable.
Solution Approach 2:
The formation control problem is segmented into local sub-problems where each USV only needs to consider its immediate neighbors and local formation requirements. This segmentation allows the system to achieve global formation flexibility through composition of simple local decisions, rather than requiring a single complex global controller.
2Reliability
If complex motion controllers are designed for collision avoidance, then safety is improved, but decision-making speed decreases
Solution Approach 1:
The decision-making network applies collision avoidance actions selectively based on local risk assessment. When collision risk is detected through local sensing, the network generates avoidance maneuvers; otherwise, it focuses on formation maintenance. This partial action approach ensures safety when needed while maintaining fast decision-making speed by avoiding unnecessary complex computations during normal operation.
3Productivity
If centralized control is used for large-scale USV swarms, then coordination is achieved, but communication overhead and control complexity increase
Solution Approach 1:
The centralized control function is segmented and distributed to individual USVs through the decision-making network. Each USV independently performs coordination tasks based on local information, eliminating the need for a complex centralized controller while maintaining swarm coordination efficiency through emergent collective behavior.
4Extent of automation
If existing formation control methods are applied to large-scale swarms (>30 vessels), then basic formation is maintained, but autonomous and collaborative capabilities are limited
Solution Approach 1:
The decision-making network enables each USV to autonomously perceive its environment, make decisions, and execute actions independently. This self-service capability scales naturally with swarm size, as each additional USV contributes its own autonomous capabilities without requiring proportional increases in centralized control complexity, thereby enhancing both autonomy and collaborative capability in large-scale swarms.
Data Source
AI summary
The present disclosure discloses a method and system for formation control for an USV swarm via a CEDRL. The method includes: designing a desired formation pattern based on a formation hierarchical virtual leader strategy, establishing an USV desired location library, and assigning a location index to a desired location of each USV in a formation; updating the desired location of each USV and the corresponding location index via an USV formation local consensus strategy in a case where there is a risk of collision between USVs; and acquiring an actual geolocation of each USV in real time, and adopting a surface vessel control decision-making network to direct the USV toward a latest desired location. An autonomous collaborative formation of a large-scale USV swarm may be realized by the present disclosure.