Trip Modelling and Optimisation equation for GTP (Goods to People)

  1. Trip Modelling
    In typical GTP (Goods to People) the trip will define as follows:
    a) AGV picks the inventory carrier (Tote or Rack)
    b) Brings the inventory carrier to the Pick Station
    c) Waits for the operator to pick the item (This time will vary based on number of items to be picked from the inventory carrier)
    d) Keep the inventory carrier back. This is when the AGV will be free to perform next order

The current API takes into account only one way that is pickup and delivery. The AGV will be assumed available when it reaches the delivery point. Is there a way to model this kind of round trip.

  1. Optimisation equation
    In case of GTP the optimisation has to be done w.r.t. the Pick operators idle time. The AGVs should reach the Pick station just in time so that the Pick operator has work to do. The current objective are minimising things like AGVs or Path Cost etc. With the round trip modelling, optimising the pick operator work time will be another requirement for GTP kind of use case.

Thank you for your question.
If I understand you correctly, it sounds like each AGV picks 1 inventory carrier, delivers it to the operator, waits for the operator to process the order, and then moves to the next operator (or potentially back to refill inventory).

I think you are correct in assuming pickup and delivery is the best modeling approach for the general case. To model waiting time I would suggest adding time windows (importantly service times) to each order.

  • The earliest pickup time for the inventory carrier would be when it is full with the required inventory. The latest time could be set at your discretion. The service time for picking up the carrier is the time required for the AGV to prepare the inventory carrier for transit (time spent at the inventory carrier location while not moving towards the operator)

  • The earliest time and latest time for delivering the carrier to the operator can be set at your discretion. Importantly, the service time should be set to the estimated pick time for that order.

With regard to the optimization objective, if you set your cost matrix to represent time (rather than distance) between each pair of locations you will now minimize total time for all orders, and should therefor also be minimizing the idle time for the operators.

Additional consideration: If your inventory carriers all start and end in approximately the same location and are full and ready to go when you start the optimization problem, you could also try modeling this as a vehicle routing problem with time windows (vrptw) problem were the start and end location for the inventory carriers is treated as the depot. Time windows could be used similarly here.