Numerical mixed-integer optimal control and motorized traveling salesmen problems

Oskar von Stryk, Markus Glocker

2001 · 20 citations · 7 references

Concepts

Abstract

A general approach for the numerical solution of hybrid, mixed-integer optimal con- trol problems is presented. In an outer level iteration a branch-and-bound procedure is ap- plied to search the entire feasible discrete variable space. An inner level iteration contains for each actual value of the discrete variable a continuous nonlinear optimal control problem with its nonlinear dynamics defined in multiple phases and phase transitions occuring at unknown switching points (events) which must be solved numerically subject to nonlinear constraints. For this purpose, a robust and efficient direct collocation method is employed that parameter- izes both the continuous state and control variables and exploits the sparse structure in the resulting nonlinearly constrained optimization problems. The proposed approach is success- fully applied to two new hybrid optimal control benchmark problems for a motorized traveling salesman and for a team of two cooperating, motorized salesmen. R ´ ESUM´ E: Cet article pr´ esente une mg´´

References

7