Lattice-Based Trajectory Planner for Autonomous Surface Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing trajectory planning algorithms for autonomous water surface vehicles (USVs) fail to safely navigate through congested marine environments due to limitations in predicting and responding to unpredictable behaviors of civilian vessels, leading to inefficiencies and increased collision risks, especially in dynamic scenarios where vessels have different dynamic characteristics and interpretations of collision avoidance rules.
Innovation Solution
A lattice-based 5D trajectory planner that estimates collision risk and incorporates contingency maneuvers to dynamically scale control action primitives based on obstacle distribution and congestion, allowing for real-time adaptation and efficient navigation through complex environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional lattice-based deliberative planners are used to find global trajectories, then collision avoidance capability is improved, but computational cost increases significantly reducing replanning rate
Solution Approach 1:
The travel space is partitioned into multiple regions, and the state space is divided into discrete 5D states. This segmentation allows the planner to process complex navigation problems in smaller, manageable chunks, reducing computational burden while maintaining collision avoidance capability through systematic exploration of divided state spaces.
Solution Approach 2:
The planner dynamically scales control action primitives based on the distribution and concentration of moving obstacles in different regions. This dynamic adaptation allows the system to adjust its behavior in real-time, increasing replanning rate by focusing computational resources on high-risk areas while maintaining safety through continuous state space evaluation.
2Ease of operation
If control action primitives are uniformly applied across the entire travel space, then implementation simplicity is maintained, but navigation efficiency decreases in congested regions
Solution Approach 1:
The planner applies different scaling factors to control action primitives based on local obstacle density in different regions. In congested areas, control actions are scaled down to allow more conservative navigation, while in open areas, full-scale actions are permitted. This local differentiation improves navigation efficiency without significantly complicating the overall system implementation.
Solution Approach 2:
The planner combines lattice-based heuristic search with dynamic scaling of control action primitives into a unified framework. By merging these two approaches, the system maintains the structural simplicity of lattice planning while incorporating the efficiency benefits of adaptive control scaling, achieving both ease of operation and improved navigation efficiency.
3Reliability
If the USV yields to civilian vessels following COLREGS, then rule compliance is improved, but navigation time increases and channel blocking occurs
Solution Approach 1:
The planner performs preliminary evaluation of multiple possible maneuvers, including both COLREGS-compliant yielding actions and potential breach actions. By pre-assessing the consequences of different choices and predicting civilian vessel responses, the system can select the most efficient path while maintaining rule compliance, reducing navigation time without sacrificing reliability.
Solution Approach 2:
The planner incorporates feedback mechanisms that evaluate the effectiveness of yielding maneuvers in real-time. By monitoring whether yielding actions actually prevent collisions and block channels, the system can adjust its behavior to minimize navigation time while maintaining COLREGS compliance, optimizing the balance between rule following and efficient passage.
Data Source
AI summary
A planning module for a water surface vehicle can determine a vehicle trajectory that avoids one or more moving obstacles, such as civilian marine vessels, by performing a lattice-based heuristic search of a state space for the surface vehicle and selecting control action primitives from a predetermined set of control action primitives based on the search. The planning module can separate a travel space into a plurality of regions and can independently scale the control action primitives in each region based on the moving obstacles therein. The heuristic search includes evaluating a cost function at each state of the state space. The cost function can be based on at least predicted movement of the obstacles responsive to respective maneuvers performed by the surface vehicle at each node of the search.