The capability of avoiding collisions in an environment with obstacles is a necessary function for an autonomous unmanned surface vehicle (USV).
We propose a novel multiple obstacles avoidance approach for USVs called the three-dimensional multiple obstacle avoidance (TDMOA) algorithm. This version releases the core implementation for TDMOA, containing seven following modules:
- World model module, built from electronic charts and the dynamic local environment, can provide detailed local marine information for other down-stream modules;
- Navigation controller module, responsible for velocity control, course control and trajectory tracking, can steer the USV to track the planned trajectory;
- Prediction module, assuming that obstacles proceed at their current statuses, can predicte obstacle's statuses (like speed, heading and position) with timestamp in a future time;
- COLREGs-based risk domain model module, inspired by the ship safety domain concept, can quantify the threat/rule cost with maritime rules;
- 3D risk model module, introducing the time dimension to the marine plane, can establish a 3D risk model with the predictive risk domains and represent the local movement trend of multiple obstacles and potential hazards to the USV in the future time.
- Path search module, a improved D star lite algorithm, can search an optimal local obstacle avoidance path in cycles;
- Simulation module, integrated with a custom autonomy suite, can validate the feasibility and effectiveness of our proposed approach in tackling the multi-obstacle avoidance problem;
Vessel Navigation Controller Module consists of Vessel model, USV model and USV Navigation module.
The navigation controller module is strongly related to the motion dynamics of a specific USV. We use the real trial data to build the Motion Model for our vessel.
You need to update the Motion Model if applying to your specific USV or vessels.
In the directory, run
python3 VesselModel/USVNaviControl.py
You will get the demo for the navigation controller, as shown below.
