Robust Execution of BDI Agent Programs by Exploiting Synergies Between Intentions
Authors: Yuan Yao, Brian Logan, John Thangarajah
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We present the results of a preliminary empirical evaluation of SP in a range of scenarios of increasing difficulty. The results suggest our approach out-performs existing failure handling mechanisms used by state-of-the-art BDI languages. |
| Researcher Affiliation | Academia | Yuan Yao School of Computer Science University of Nottingham Nottingham, UK yvy@cs.nott.ac.uk Brian Logan School of Computer Science University of Nottingham Nottingham, UK bsl@cs.nott.ac.uk John Thangarajah School of Computer Science and IT RMIT University Melbourne, Australia john.thangarajah@rmit.edu.au |
| Pseudocode | Yes | The pseudocode for the scheduler is shown in Algorithm 1. |
| Open Source Code | No | The paper does not provide concrete access to its source code, nor does it state that the code is available in supplementary materials or a public repository. |
| Open Datasets | No | The paper states: our evaluation is based on sets of randomly-generated, synthetic goal-plan trees representing the current intentions of an agent in a simple static environment. It does not use a pre-existing publicly available dataset and provides no access information for the generated data. |
| Dataset Splits | No | The paper describes generating synthetic data for evaluation ('randomly-generated, synthetic goal-plan trees') but does not specify traditional train/validation/test splits, percentages, or absolute sample counts for data partitioning. The evaluation is performed on 'trials' with generated data. |
| Hardware Specification | No | The paper states that 'SP requires 35 milliseconds to return the action to be executed at this deliberation cycle' which implies computation, but it does not provide any specific hardware details such as GPU/CPU models, memory, or cloud instance types used for running experiments. |
| Software Dependencies | No | The paper mentions BDI agent programming languages like JACK, Jason, and Jadex, but it does not specify any software dependencies (libraries, frameworks, or solvers) with version numbers that would be required to replicate their implementation or experiments. |
| Experiment Setup | Yes | SP was configured to perform 1000 iterations (α = 1000). The paper also describes parameters for generating the experimental scenarios, such as the depth of the tree, plan branching factor, goal branching factor, maximum number of actions in a plan, probability of fallible action, and number of environment variables. |