Adaptive Routing

Adaptive routing for physical delivery services

1. Optimizes emergency response times

2. We have designed a decentralized reinforcement learning system and trained it using real emergency response call records

3. Decentralized systems are close to centrally discovered optimal solutions under a large range of conditions

4. Policies learned within decentralized systems are much more interpretable and explainable

5. Reinforcement Learning policies need to be trained on historical spatiotemporally indexed demand data for your application

6. System needs to be integrated with your dispatch or routing service

7. Needs brief pilot testing to establish real-world calibration of parameters