Control and Planning for Nautical Vehicles

Project Description

Command and Trajectory Planning for Nautical Vehicles

Nautical vehicles are difficult to control precisely due to multiple sources of disturbances (currents and winds) and a propulsion system with several constraints. Several modern control and planning methods (e.g., MPC or reinforcement learning) promise to work well in this type of environment. This project focuses on developing the most robust control algorithm possible against disturbances, exploring various modern methods adapted to the nautical context. The student working on this project will develop algorithms (control law, trajectory optimization, dynamic simulation, and/or machine learning), conduct tests in a realistic simulation environment including a driving simulator, and participate in trials with full-scale vehicles.

About the research project

BRP, CM labs Simulations, LeddarTech, Université de Sherbrooke and McGill University will co-develop connected and perceptive vehicles for advanced research purposes. The collaboration will also involve Centre de Technologies Avancées - BRP - UdeS (CTA). This joint initiative will enable the parties to investigate, test and validate advanced driver assistance system (ADAS) features such as collision avoidance in off-road environments. Prototype vehicles will be able to "see" their environment in real-time with perceptive sensors, including LiDAR sensors provided by LeddarTech, radars, and cameras while also having the ability to communicate with one another using the latest vehicle-to-everything (V2X) protocols. Vehicles will incorporate some Level II and Level III driving capabilities to enhance driver/passenger safety. To accelerate the development, a digital twin will be created using CM Labs' Vortex Studio. The advanced digital twin will be used by students and professionals as a software-in-the-loop and driver-in-the-loop sandbox. The vehicle will be meticulously modelled with a state-of-the-art high-fidelity simulation that will include the perceptive sensors and validated real-time tire-soil interaction models, thus enabling the team to conduct virtual validation plans, train neural networks and generate large amounts of virtual data.

Funding offered

Yes

Partner(s)

Bombardier Produits récréatifs (BRP) , CM Labs Simulations Inc.

The last update was on 9 July 2024. The University reserves the right to modify its projects without notice.