Field report
Try a sample report
Training
Plays simulated flood scenarios in the OpenEnv environment. Every run is scored; the bandit and the lesson memory update. rl_policy is a neural policy trained offline with reinforcement learning (PyTorch).
Pipeline
1Understand
2Plan
3Dispatch
4Disruption
5Self-heal
6Score
7Learn
Operations map
Depot
Open request
Field report
Truck leg
Detour
Blocked
Flooded
Event log
No activity yet. Choose a sample report and dispatch, or run training.
Performance
–Last score
–Average, last 5
0Episodes
–Change vs first 5
Score = share of the best possible outcome. Dots: each episode. Line: 5-episode average.
Policy · bandit
Mostly uses the best-scoring strategy, tries the others 20% of the time.