Multi-USV Collaborative Search Using DDQN and Probability Grids
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current unmanned surface vessel (USV) systems face challenges with low search efficiency and accuracy due to limited sensing ranges and the need for manual intervention, which can be hazardous and inefficient.
Innovation Solution
A method and system for controlling multi-USV collaborative search, utilizing a double deep Q-network (DDQN) network structure and probability graph mode to determine task regions, environmental perception, and execution actions for each USV, enabling real-time target search strategy optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If multiple USVs are deployed to cover a large task region, then the search coverage area increases, but the coordination and task allocation complexity increases
Solution Approach 1:
The patent divides the large task region into multiple sub-regions and assigns different USVs to search specific sub-regions. The search area is segmented into grid cells, and each USV is responsible for searching specific cells based on target existence probabilities, thereby reducing coordination complexity while maintaining comprehensive coverage.
Solution Approach 2:
The patent introduces a probability dimension by calculating target existence probabilities for different grid cells. This probabilistic approach adds a new dimension to the search strategy, allowing the system to prioritize high-probability areas and optimize the allocation of multiple USVs across the search region.
2Reliability
If the USV team operates autonomously without manual intervention, then the operational safety improves, but the system complexity increases
Solution Approach 1:
The patent implements autonomous operation where the USV team performs environmental perception, target detection, and task execution without continuous manual intervention. The system uses automated algorithms for probability calculation, task allocation, and coordination, enabling self-service operation that improves safety while managing complexity through standardized autonomous protocols.
Solution Approach 2:
The patent incorporates feedback mechanisms where the USVs continuously perceive environmental information, update target existence probabilities, and adjust their search strategies based on real-time data. This feedback loop enables autonomous adaptation to changing conditions while maintaining operational safety through automated decision-making.
3Length of stationary object
If high-power observation sensors are equipped on USVs, then the sensing range increases, but the device complexity and cost increase
Solution Approach 1:
The patent combines the sensing capabilities of multiple USVs to achieve comprehensive coverage of the search region. By coordinating the sensors of multiple vessels, the system effectively extends the sensing range without requiring individual USVs to carry high-power sensors, thereby avoiding the associated complexity and cost increases.
Solution Approach 2:
The patent introduces a probabilistic dimension to compensate for limited sensing range. By calculating target existence probabilities based on environmental perception and search history, the system can prioritize high-probability areas within the limited sensing range of each USV, effectively extending the operational search capability without requiring high-power sensors.
4Productivity
If a single USV performs the search task, then the system simplicity is maintained, but the search efficiency decreases
Solution Approach 1:
The patent segments the search task by dividing the task region into grid cells and assigning different USVs to search specific cells or sub-regions. This segmentation enables parallel search operations that significantly improve search efficiency compared to a single USV, while the standardized segmentation approach keeps coordination complexity manageable.
Solution Approach 2:
The patent implements a probabilistic search strategy where USVs focus on grid cells with higher target existence probabilities. This partial action approach, where not all cells are searched with equal intensity, improves search efficiency by concentrating resources on high-probability areas while reducing the overall coordination burden compared to exhaustive search methods.
Data Source
AI summary
A method and system for controlling multi-unmanned surface vessel (USV) collaborative search are disclosed which relate to the technical field of the marine intelligent USV collaborative operation. The method includes determining a task region of a USV team; determining environmental perception information corresponding to each of the USVs at the current moment according to the task region and the probability graph mode; inputting the environmental perception information corresponding to each of the USVs at the current moment into the corresponding target search strategy output model respectively to obtain an execution action of each of the USVs at the next moment; sending the execution action of each of the USVs at the next moment to a corresponding USV execution structure to search for underwater targets within the task region. The target search strategy output model is obtained by training based on a training sample and a DDQN network structure.