Software

Temporal Relaxation Benchmarks

These benchmarks were created based on the following two papers:

Continuously Relaxing Over-constrained Conditional Temporal Problems through Generalized Conflict Learning and Resolution, Peng Yu and Brian Williams, the Twenty-third International Joint Conference on Artificial Intelligence (IJCAI-2013), Beijing, 2013

and

Resolving Uncontrollable Conditional Temporal Problems using Continuous Relaxations, Peng Yu and Cheng Fang and Brian Williams, In Proceedings of the Twenty-fourth International Conference on Automated Planning and Scheduling (ICAPS-2014), Portsmouth, 2014.

You may find copies of these papers in the publication page.

These problems are (over-constrained) conditional temporal problems with set-bounded uncertain duration, which was designed for the Conflict-directed Relaxation with Uncertainty algorithm. With some modifications these benchmarks can also be used as temporal optimization problems, such as the maximum flexibility problems (see Optimising Bounds in Simple Temporal Networks with Uncertainty under Dynamic Controllability Constraints, Jing Cui, Peng Yu, Cheng Fang, Patrik Haslum and Brian Williams, ICAPS-15).

The benchmark tests and a visualizer can be downloaded from the groups’ Dropbox folder. You may also find the setup instruction (CDRU Manual.docx) in the folder.

The slides presented at IJCAI-13 can be downloaded from here. It provides an overview of the approach used by BCDR (an earlier version of CDRU based on consistency) and the design of these benchmarks. Another set of slides prepared for ICAPS-14 explains how CDRU handles the relaxation/tightening of problems with uncertain duration, and its applications to several transportation problems (link).

Finally, you may find some example problems in the “Examples” folder. They are constructed from the domains of transit, daily activity, and deep sea exploration planning. The benchmark test cases are in the ‘RelaxationTests-NoChoice’ and ‘RelaxationTests-Withchoice’ folders. An explanation of the file format can be found in the word document ‘File format Manual.docx’. To visualize the test cases, you may use the java executable inside the folder (for *.cctp files), or the online editor at http://bicycle.csail.mit.edu/tpneditor (for *.tpn files).

We appreciate your efforts in trying out these benchmarks and any feedback is greatly appreciated. Please let us know if you have any questions. This visualizer software may not be redistributed, and can only be retained and used with the explicit written consent of the authors, Peng Yu and Brian Williams. This software is made available as is; no guarantee is provided with respect to performance or correct behavior. This software may only be used for non-commercial, non-profit, research activities. Copyright (c) 2013-2015 Peng Yu and Brian Williams, MIT.