Mixed Discrete-Continuous Planning with Convex Optimization

Authors: Enrique Fernandez-Gonzalez, Erez Karpas, Brian Williams

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental Results To showcase the new capabilities of our planner and to show that our optimization framework is fast and scalable, we present three new expressive domains and benchmark our planner against them. Since no other planner can solve these domains, we also provide a simplified, linear version of some of these domains that we use to compare our planner to Scotty and POPCORN. We compare against these planners since they support control variables, which are essential for these domains. ... Table 1: Results of Empirical Evaluation. t: Planning time in seconds; L: Plan length; S: Number of nodes expanded; N: Number of optimization problems solved; T: Mean solving time for each optimization problem in milliseconds.
Researcher Affiliation Academia Enrique Fern´andez-Gonz´alez MIT CSAIL Cambridge, MA efernan@mit.edu Erez Karpas Technion Haifa, Israel karpase@technion.ac.il Brian Williams MIT CSAIL Cambridge, MA williams@mit.edu
Pseudocode No The paper describes the algorithms and methods in prose but does not include any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not include any statements about releasing code or links to a source code repository.
Open Datasets No The paper introduces 'three new expressive domains' and a 'simplified, linear version of some of these domains' (AUV domain, ROV domain, Air Refueling domain) but does not provide concrete access information (link, DOI, formal citation) for these datasets.
Dataset Splits No The paper does not provide specific dataset split information (exact percentages, sample counts, or detailed splitting methodology) for training, validation, or testing.
Hardware Specification Yes Results: We benchmarked our planner on an Intel Core i7-3770 3.40 GHz with the Gurobi 7.0.1 solver.
Software Dependencies Yes We benchmarked our planner on an Intel Core i7-3770 3.40 GHz with the Gurobi 7.0.1 solver. ... We hypothesize that this is due to the superior performance of the Gurobi solver compared to the solvers used by Scotty (CPLEX 12.4) and POPCORN (lpsolve 5.5).
Experiment Setup No The paper describes the overall approach, including the heuristic forward search and convex optimization, and details the encoding of constraints. However, it does not specify concrete experimental setup details such as hyperparameters, learning rates, or specific adjustable training/search configurations.