Autonomous Navigation System for Maneuverable Platforms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous navigation systems for maneuverable platforms, such as unmanned vehicles, face challenges in effectively avoiding collisions and adhering to traffic regulations, especially in complex scenarios with multiple obstacles and ambiguous rules, which can lead to collisions and mission failures.
Innovation Solution
An autonomous navigation system that includes a situation awareness module for data collection from sensors, a decision module for choosing optimal course and speed based on mission objectives and obstacle avoidance, and a platform control module for executing these decisions, using a multi-objective optimization approach to weigh objectives like collision avoidance, traffic rules, and mission completion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If piecewise-defined interval programming functions are used to represent each COLREGS rule with about 600 linear pieces per rule, then the autonomous vehicle can evaluate multiple traffic rules simultaneously, but the computational complexity becomes unwieldy with 600^5+1 nodes in the decision tree
Solution Approach 1:
The patent segments the complex multi-objective optimization problem into separate single-objective optimization problems. Instead of evaluating all 600^5+1 combinations simultaneously, the system performs k separate optimizations, each focusing on one objective function with its own set of 600 linear pieces. This reduces computational complexity from evaluating the entire decision tree to evaluating k separate, manageable optimization problems.
Solution Approach 2:
The patent transforms the high-dimensional decision space problem into a series of lower-dimensional problems by introducing a new dimension for objective weighting. Instead of searching through the massive 600^5+1 node tree, the system adds an objective weight dimension and performs optimizations along this new dimension, effectively reducing the computational burden while maintaining solution quality.
2Measurement precision
If the system evaluates all possible combinations of multiple objectives with 600 linear pieces each, then the most optimal solution can be found, but the computational time and resources become prohibitively large
Solution Approach 1:
The patent divides the exhaustive search problem into k separate optimization tasks. Each task evaluates one objective function independently with its 600 linear pieces, requiring minimal computational time. The final optimal solution is derived by combining results from these k separate evaluations according to objective weights, achieving high precision without the prohibitive time cost of evaluating all combinations simultaneously.
Solution Approach 2:
The patent performs preliminary evaluations of each objective function separately before combining them. By pre-evaluating each of the k objective functions with their respective 600 linear pieces independently, the system prepares optimized sub-solutions that can be quickly combined using objective weights, avoiding the need to evaluate all possible combinations in one exhaustive search.
3Reliability
If the autonomous platform continuously monitors and adjusts course and speed to avoid multiple moving obstacles, then collision avoidance improves, but the system becomes less adaptable when faced with ambiguous or contradicting traffic regulations
Solution Approach 1:
The patent changes the parameter representation from fixed rule-based decisions to dynamic objective weighting. Instead of rigidly following ambiguous COLREGS rules, the system represents traffic regulations as flexible objective functions with adjustable weights. This allows the platform to adapt to contradictory or ambiguous situations by dynamically adjusting objective weights based on the specific context, maintaining collision avoidance while improving adaptability.
Solution Approach 2:
The patent introduces dynamics into the decision-making process by making objective weights adjustable rather than fixed. The system can dynamically modify the importance of different traffic rules and objectives based on the current situation, allowing it to adapt to ambiguous or contradicting regulations while continuously monitoring obstacle positions to maintain collision avoidance reliability.
Data Source
AI summary
An autonomous navigation system and method for a maneuverable platform are provided herein. The method may include: obtaining a plurality of objectives relating to a maneuverable platform; determining a plurality of options of direction and speed for the platform; autonomously selecting, one option of the plurality of options, in order to achieve said objectives, by calculating a weighted grade of each option based on the weights of the objectives and said grading scheme; and periodically repeating: the receiving with updated platform and obstacles data and the autonomously selecting with the updated platform and obstacles data, wherein the autonomously selecting is executed by a computer processor and includes, for each obstacle, calculating projected positions of the platform and of that obstacle to determine the distance between the platform and the obstacle at the closest point of approach and estimated time to arrive to the closest point of approach.