Automated Scenario Generation for Collision Avoidance Stress Testing
Introduction:
Ensuring the reliability and effectiveness of collision avoidance algorithms is paramount for developing safe autonomous ships. The COLAV simulator provides a robust platform for testing such algorithms using predefined, custom, or random scenarios. However, manual scenario creation can be time-consuming and may not cover all potential scenarios. This proposal aims to develop an intelligent algorithm to automatically generate or adapt scenarios within the COLAV simulator to stress-test collision avoidance algorithms effectively.
Objectives:
The primary objective of this specialization/master project is to develop an algorithm that can automatically create or adapt scenarios in the COLAV simulator to identify scenarios where collision avoidance algorithms exhibit unsatisfactory performance.
Proposed Steps:
- Literature Review: Conduct a comprehensive review of existing literature on automatic stress testing methods and scenario generation techniques, focusing on their applicability to collision avoidance algorithms and maritime simulations.
- Familiarization with COLAV Simulator: Learn how to set up and use the COLAV simulator framework, exploring its existing scenario creation and evaluation tools.
- Algorithm Development: Design and implement an intelligent algorithm capable of automatically generating or adapting scenarios within the COLAV simulator.
- Scenario Generation Testing: Validate the developed algorithm by generating a diverse set of scenarios within the COLAV simulator and assessing their effectiveness in stress-testing collision avoidance algorithms.
- Evaluation and Validation: Evaluate the performance of the developed algorithm using metrics such as scenario diversity, coverage, and effectiveness in stress-testing collision avoidance algorithms.
- Report: Document the findings, methodologies, and results into a comprehensive thesis document. Analyze the performance of the developed algorithm, discuss its implications, and propose recommendations for future research.
This specialization/master project will be conducted in collaboration with Kongsberg Maritime, a leading provider of maritime technology solutions.