DynStack Competition - Dynamic Stacking Optimization in Uncertain Environments

This 2-page extended abstract gives a short overview of the Dyn-Stack competition, which has been hosted at the Genetic and Evolutionary Computation Conference (GECCO) since 2020. The challenge is to generate optimized solutions for one of three different dynamic optimization scenarios in the domain of warehouse operations. These scenarios are simulated and each simulation model, executed in real-time, publishes its world state (i.e. all available information, e.g. existing transport orders and item stock) into message queues regularly. External optimizers can attach to these queues and send back respective optimization solutions in the form of crane schedules, which are then executed by simulated crane agents in the simulated systems. Dynamic changes within the simulations necessitate updates of crane schedules on a regular basis.