Understanding Pathmind Tools
Before you begin, it is important to understand the three tools needed to use Pathmind with AnyLogic: Pathmind Helper, Pathmind Training, and Pathmind Serving. These tools provide everything you need to go from proof-of-concept to implementing RL into your operations.
Pathmind Helper
Pathmind Helper is an AnyLogic pallette item. Drop Pathmind Helper into a model and use it to add RL functions.
Pathmind Training
Pathmind Training is a web app. Export an AnyLogic model to Pathmind Training to train RL around metrics you define.
Pathmind Serving
Pathmind Serving is a deployment service. Use Pathmind Serving to apply your trained RL policies in real-world workflows.
1. Create a Pathmind Account
A Pathmind account is required to access Pathmind Training and Pathmind Serving. A Trial account includes everything you need to complete this guide. You can upgrade your account at any time to access all of Pathmind’s features.
2. Download Pathmind Helper
Adding Pathmind Helper to AnyLogic will allow you to expose and edit RL fields directly in your models. Adding these RL functions converts your simulation data into a format that Pathmind Training can understand to train RL.
3. Start Training RL Policies
The best way to famiarize yourself with the Pathmind workflow is to dive into our tutorials. These tutorials work out of box and already have Pathmind integrated, so you can get an understanding of how to map RL in your own models. Each tutorial includes a complete AnyLogic model with RL fields already filled out in Pathmind Helper. You can export these models directly into Pathmind Training and begin training an RL policy right away.
We recommended starting with the tutorials below. Check out our full tutorials page for even more examples.
Note: If this is your first time exporting a model to Pathmind Training, you will need to link your Pathmind Account with AnyLogic using these steps.
Getting Started
Introducing Pathmind and using RL in simulations with a simple stochastic model.
Multi-Echelon Product Delivery
Comparing RL to an optimizer/heuristic hybrid with a supply chain example.
Automated Guided Vehicles
Coordinating multiple RL agents using automated guided vehicles (AGVs).
4. Deploy Trained RL Policies
Once you have a trained RL policy that meets your goals, you’ll want to apply it to you real-world operations. You can deploy trained Pathmind-trained RL using three methods: Microsoft Excel, REST API, and On-Premise.
Microsoft Excel
Include RL predictions within existing Microsoft Excel worksheets.
REST API
Integrate RL in Salesforce, SAP, or any web-based app you prefer.
On-Premise
Deploy RL offline to accomodate security, firewall, or internet restrictions.
Next Steps
- Visit the Pathmind Help Center to see our full library of docs.
- Access more of Pathmind’s features by upgrading your account.
- Contact the Pathmind team if you still have questions or want to discuss a project for your organization.