Warehouse Order Dispatching Problem

Follow next points to find and test your specific best solution (let us indicate the project root with /).

1. Define your warehouse. First of all, navigate to the /src folder and set your warehouse topology in file initial_state.py.

2. Insert your data. Then, go to /data and insert your data or generate them with the script gen_exp.py. Data have to be split in three subset (training set, validation set and test set) and each instance have to be represented with the format t_arrival,pick_zone_idx,drop_zone_idx. In the beginning of each file CSV, you must indicate the titles for all columns separated by commas.

3. Set GP hyperparameters. Set main GP hyperparameters in file /src/gp_hyperparams.py. To set more of them, you can modify directly the /src/gp_main.py file.

4. Start tensorboard service. Before training, run the following command to show real-time information about training.

   tensorboard --logdir=./runs --samples_per_plugin=text=30 
   

   If your text data seem to be subsampled, you can try to increase the last number in the command.

5. Start training. Run /src/gp_main.py script and wait the end of the GP algorithm.

6. Testing. You can test the solution found by copying it from information printed in tensorboard and paste it in the correct place in file /src/gp_test.py. The script shows the three objective values of the solution on the test set.

7. Simulation. You can also simulate the solution found running /scr/sim_main.py after setting the correct policy.

You can find the report and the presentation of the project at /docs directory.