Marine Vessel Reachable Set Calculation for Real-Time Collision Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current reachability analysis methods are computationally intensive and unable to effectively handle high-dimensional nonlinear systems, such as marine vessels with six states, making real-time implementation challenging for safe navigation and collision avoidance.
Innovation Solution
A method involving discretization of a partial differential equation based on a marine vessel model, solving a minimization problem to approximate a value function, and determining the reachable set using a splitting method and coordinate descent, which reduces computation time and provides accurate real-time predictions of vessel positions and maneuvering capabilities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional reachability analysis methods are used for marine vessels, then accuracy of reachable set calculation is maintained, but computation time becomes excessively long and real-time implementation is impossible
Solution Approach 1:
The patent applies segmentation by dividing the high-dimensional state space into multiple lower-dimensional subspaces. Instead of computing the reachable set for the entire 6-dimensional marine vessel system at once, the method partitions the state space and computes reachable sets for individual subsystems or state variables separately, then combines the results. This segmentation reduces the computational complexity from exponential in the full dimension to manageable levels in reduced dimensions, enabling real-time computation while preserving accuracy.
2Reliability
If reachability analysis is applied to high-dimensional nonlinear systems like marine vessels, then comprehensive safety assessment is achieved, but the curse of dimensionality makes the problem computationally intractable
Solution Approach 1:
The patent employs dimensionality change techniques by transforming the high-dimensional nonlinear reachability problem into a series of lower-dimensional problems. The method introduces auxiliary variables or uses coordinate transformations that map the complex 6-dimensional marine vessel state space into a framework where computation can be performed in reduced dimensions. This dimensional transformation maintains the essential safety assessment capabilities while avoiding the exponential computational burden of the curse of dimensionality.
3Productivity
If real-time reachable set calculation is implemented for marine vessels, then collision avoidance and safe navigation are enabled, but the computational requirements exceed available onboard processing capabilities
Solution Approach 1:
The patent applies partial action by computing reachable sets for critical subsystems or state variables with higher priority for collision avoidance, rather than computing all aspects of the full 6-dimensional state space with equal detail. The method identifies which dimensions of the state space most critically affect safety (e.g., position and velocity in collision-prone directions) and allocates computational resources preferentially to those dimensions. This partial computation approach achieves sufficient real-time performance with available onboard processing power while maintaining adequate safety assessment for the most critical parameters.
Data Source
Figure 1~3

AI summary
A method of determining a reachable set (R[t0, t1]) of states of a marine vessel (7), including a position of the marine vessel (7), that can be obtained by the marine vessel (7) from an initial state x0 at an initial time to, during a time range from the initial time to to a later time ti, wherein the method comprises: a) discretizing a partial differential equation which depends on a model of the marine vessel (7), the discretizing being performed based on a subdivision of the time range, b) solving a minimisation problem of the discretized partial differential equation to obtain an approximation of a value function of the partial differential equation, and c) determining the reachable set (R[t0, t1]) based on the approximation of the value function.