Berthing AI – Advisory System V3.3.0

V3.3.0 • Rule/Hybrid Risk + visualization • LSTM/GRU Training/Validation/Forecast • Safe Policy V3.2.0 • portlogistics.vn

Cases
Warning / Critical
Mean Data Quality
Mean Hybrid Risk

Berthing trajectory

Risk score

Speed / lateral speed

Data Table

Ground Truth / Label Source

No expert labels loaded. Training will use rule-derived labels as fallback.

Calibration & Group CV

Train the risk model to calculate calibration and cross-validation metrics.

Hybrid AI result status

Hybrid Risk Model has not been trained yet. Train the model and run Predict Risk to display Hybrid AI charts.

Model validation

No model trained yet.

Hybrid configuration

Rule + Random Forest + Gradient Boosting

Risk report

No model trained yet.

Trajectory validation / future forecast

V3.3.0 Backend model comparison

Train LSTM/GRU on the Python backend first.
V3.3.0: frontend sends trajectory data to Python backend → train LSTM/GRU → validate ADE/FDE/RMSE → forecast X/Y.
Ready.

Berthing Animation – Selected CaseID

Current vessel state

Select a CaseID and press Start.

Animation notes

Blue line: historical track already traveled.
Red point/triangle: current vessel position and heading.
Black vertical line: berth face (x = 0).
Dashed horizontal lines: preferred safety corridor (±30 m).
This animation is for advisory/research visualization only.

Safe policy simulation – Distance-Aware V3.2.0

Train and simulate safe policy.

System log

System initialized.
Processing...
Please wait.
0%