Dynamic Multi-Agent Task Allocation with Spatial and Temporal Constraints
Authors: Sofia Amador, Steven Okamoto, Roie Zivan
AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We empirically compare our algorithm to state-of-the-art incomplete methods, both centralized and distributed, on law enforcement problems inspired by real police logs. The results show a clear advantage for FMC TA both in total utility and in other measures commonly used by law enforcement authorities. |
| Researcher Affiliation | Academia | Sofia Amador and Steven Okamoto and Roie Zivan Industrial Engineering and Management Department Ben-Gurion University of the Negev Beer-Sheva, Israel {amador, okamotos, zivanr}@post.bgu.ac.il |
| Pseudocode | No | The paper describes the algorithms in prose within the "FMC-Based Task Allocation" section (Centralized Algorithm, Distributed Algorithm) but does not provide structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described. |
| Open Datasets | No | The paper states that problems were generated "based on real police logs" and "Event types were selected randomly according to the distribution of real event types provided by law enforcement authorities in our home city," but does not provide concrete access information (link, DOI, citation) to a publicly available or open dataset. |
| Dataset Splits | No | The paper describes generating problems based on police logs for simulation but does not specify exact training/validation/test dataset splits or cross-validation setup, as would be typical for a fixed dataset. |
| Hardware Specification | No | The paper does not provide specific hardware details (e.g., GPU/CPU models, memory amounts, or cloud instance types) 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 | The city is represented by a rectangular region of the Euclidean plane of size 6 x 6 kilometers, divided into 9 neighborhoods of size 2 x 2, each with a patrol task. We simulated 8-hour shifts as in real police departments, with 9 agents patrolling (one in each neighborhood) at the beginning of each shift. The number of tasks arriving (i.e., the load) could vary between shifts. Tasks arrived at a fixed rate and were distributed uniformly at random in the city. There were four types of events of decreasing importance I(v) = 2400, 1600, 1200, 800 from type 1 to type 4, respectively. Patrols had I(v) = 500. Event types were selected randomly according to the distribution of real event types provided by law enforcement authorities in our home city: 30%, 40%, 15%, 15% of events were of type 1 to 4, respectively. Based on estimates provided by the police, the workloads were drawn from exponential distributions with means 58, 55, 45, 37 for events of type 1 to 4, respectively. We assumed Cap improved quality of task execution for each agent up to a maximum number of agents Qv: Cap(v, q) = min{ q / Qv I(v), I(v)} where Qv = 3, 2, 1, 1 for tasks of type 1 to 4, respectively. The discount function used was δ(v, t) = 0.9t for all v. |