Faster Conflict Generation for Dynamic Controllability

Authors: Nikhil Bhargava, Tiago Vaquero, Brian Williams

IJCAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 5 Empirical Results To understand the performance of our algorithm, we randomly generated a series of STNUs to model the parallel deployment of autonomous underwater vehicles (AUV). Each of the AUVs had to navigate to a series of different sites and conduct experiments along the way; the travel durations were assumed to be contingent links while the AUVs were given the agency to control how long the experiments would take. For all trials we assumed 70 AUVs, which each had 70 tasks to complete. In our trials, we intentionally overconstrained the STNUs to make them uncontrollable. This helped us understand how our algorithm performed under a true plan relaxation task. Over the course of 50 trials, we took our randomly generated STNU and ran it through the dynamic controllability checking algorithm. Based on the returned conflict, we relaxed a constraint to eliminate the conflict. We repeated these iterations 10 times each for the incremental and non-incremental dynamic controllability algorithm. From Figure 2, we see that the time it takes for successive runs of the incremental algorithm tend to decrease over time whereas successive runs of the non-incremental algorithm increase. This matches quite closely to our expectations the non-incremental algorithm operates by terminating as soon as it finds a semi-reducible negative cycle.
Researcher Affiliation Academia Nikhil Bhargava, Tiago Vaquero, Brian Williams Massachusetts Institute of Technology {nkb, tvaquero, williams}@mit.edu
Pseudocode Yes Algorithm 1: Dynamic Controllability algorithm that reports conflicts
Open Source Code No The paper does not include any statement about releasing source code or provide any links to a code repository.
Open Datasets No To understand the performance of our algorithm, we randomly generated a series of STNUs to model the parallel deployment of autonomous underwater vehicles (AUV).
Dataset Splits No The paper does not provide specific dataset split information (percentages, sample counts, or citations to predefined splits) as it uses randomly generated STNUs for each trial instead of a fixed dataset with splits.
Hardware Specification No The paper does not provide any specific hardware details such as CPU/GPU models, processor types, or memory used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiment.
Experiment Setup Yes For all trials we assumed 70 AUVs, which each had 70 tasks to complete. In our trials, we intentionally overconstrained the STNUs to make them uncontrollable. This helped us understand how our algorithm performed under a true plan relaxation task. Over the course of 50 trials, we took our randomly generated STNU and ran it through the dynamic controllability checking algorithm. Based on the returned conflict, we relaxed a constraint to eliminate the conflict. We repeated these iterations 10 times each for the incremental and non-incremental dynamic controllability algorithm.