Table 9 Test scenario specifications.

From: Deep reinforcement learning model for Multi-Ship collision avoidance decision making design implementation and performance analysis

Scenario

Description

Vessels

Initial Configuration

Challenge Level

1

Head-on encounter

2

Two vessels on reciprocal courses

Basic COLREGs Rule 14 situation

2

Crossing situation

3

Three vessels in crossing paths with defined give-way relationships

Intermediate - multiple COLREGs rules apply

3

Overtaking

3

Three vessels traveling same direction at different speeds

Intermediate - speed management critical

4

Complex encounter

5

Mixed crossing and head-on situations

Advanced - multiple simultaneous interactions

5

High-density traffic

10

Congested waterway with vessels on various courses

Expert - complex prioritization needed

  1. Each scenario was tested under three environmental conditions: (1) Ideal conditions (no external disturbances) (2) Moderate disturbances (random noise added to state transitions, σ = 0.1) (3) Severe disturbances (random noise added to state transitions, σ = 0.3).