Abstract:Consider optimizing a periodic schedule for an automated production plant as a last step of a more comprehensive design process. In our scenario, each robot’s cyclic sequence of operations and trajectories between potential waiting points have already been fully specified. Further given are those precedences that fix sequence requirements on operations between different robots. It remains to determine the starting time for each operation or movement of each robot within a common cyclic time period so as to avoid collisions of robots that operate in the same space simultaneously. So the task is to find a conflict-resolving schedule that minimizes this common periodic cycle time while observing all precedence relations and collision avoidance constraints. The proposed cycle time minimization problem for robot coordination has, to the best of our knowledge, not been studied before. We develop an approach for solving it by employing binary search for determining the smallest feasible period time of an iso-periodic event scheduling problem (IPESP). This is a variant of the periodic event scheduling problem in which the objects that have to be scheduled need to obey exactly the same period time. The possibility to wait arbitrarily long at waiting points turns out to be essential to justify the use of binary search for identifying the minimum cycle time, thereby avoiding bilinear mixed integer formulations. Special properties of the given scenario admit bounds on the periodic tension variables of an integer programming formulation. Although the IPESP subproblems remain NP-complete in general, these bounds allow solving real-world instances sufficiently fast for the approach to be applicable in practice. Numerical experiments on real-world and randomly generated data are supplied to illustrate the potential and limitations of this approach. Summary of Contribution: When designing automated production plants, a crucial step is to identify the smallest possible per unit period time for the production processes. Based on periodic event scheduling ideas, we develop and analyze mathematical methods for this purpose. We show that the algorithmic implementation of our approach provides an answer to current real-world designs in reasonable time.

The Tropical and Zonotopal Geometry of Periodic Timetables

The tropical and zonotopal geometry of periodic timetables

On the Split Closure of the Periodic Timetabling Polytope

Symmetry for Periodic Railway Timetables

Finding Robust Periodic Timetables by Integrating Delay Management

Periodic railway timetabling with event flexibility

New Perspectives in Scheduling Theory

Periodic Event Scheduling for Automated Production Systems

Periodic Scheduling and Packing Problems

Packing-Inspired Algorithms for Periodic Scheduling Problems with Harmonic Periods

Period integrals of hypersurfaces via tropical geometry

Solving the Periodic Timetabling Problem using a Genetic Algorithm

Converging Periodic Boundary Conditions and Detection of Topological Gaps on Regular Hyperbolic Tessellations

Extremum Cycle Times in Time Interval Models

Designing alternative railway timetables under infrastructure maintenance possessions

Timetable problem using constraint programming

A Tropical Geometric Approach To Exceptional Points

Semidefinite Programming in Timetabling and Mutual-Exclusion Scheduling

A combinatorial extension of tropical cycles

Periodic lozenge tilings of the plane

Toric configuration spaces: the bipermutahedron and beyond