Hybrid Cost Automata Applied to Energy Optimization

Abstract

For optimization of hybrid systems, the ordinary hybrid automatonis extended to include a local cost criterion in each location of thehybrid model. The extended model is called a hybrid cost automaton,where generally each local cost depends on the continuoustimestate trajectory in corresponding location, and each cost isupdated as long as its location is active. The total cost criterionis achieved by adding the individual costs from each active location,where the best route among alternative locations is selected inthe optimization. This model is used to achieve an energy optimalschedule for robot cells. Each robot operation including its energycost is then represented as a location, and a number of interactingrobots are modeled as a set of hybrid cost automata, one for eachindividual robot. The energy consumption for each robot operationis modelled and parameterized as a function of the operation’s executiontime, and the energy-optimal schedule is derived by solvinga mixed-integer nonlinear programming (MINLP) problem.