SCRAM: Scalable Collision-avoiding Role Assignment with Minimal-Makespan for Formational Positioning
Authors: Patrick MacAlpine, Eric Price, Peter Stone
AAAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Primary contributions of this paper include a complete specification of SCRAM, the presentation of role assignment functions for assigning agents to target positions, algorithms (both new and existing) for computing the role assignment functions,1 as well as a thorough theoretical and empirical analysis of the role assignment problem, with application to the Robo Cup robot soccer domain and potentially far beyond. |
| Researcher Affiliation | Academia | Patrick Mac Alpine and Eric Price and Peter Stone Department of Computer Science The University of Texas at Austin Austin, TX 78701, USA {patmac, ecprice, pstone}@cs.utexas.edu |
| Pseudocode | Yes | Algorithm 1 MMDR O(n5) Polynomial Time Impl. Input: Agents := {a1, ..., an}; Positions := {p1, ..., pn} Edges := {a1p1, a1p2, ..., anpn}; |aipj| := euclidean Dist(ai,pj) |
| Open Source Code | Yes | Videos of SCRAM role assignment in action, as well as C++ implementations of the role assignment algorithms, can be found at http://www.cs.utexas.edu/ Austin Villa/sim/3dsimulation/ Austin Villa3DSimulation Files/2013/html/scram.html |
| Open Datasets | No | The paper describes using Robo Cup 2D and 3D simulation environments and generating mapping scenarios, but does not provide specific access information (link, DOI, formal citation) for a publicly available dataset. |
| Dataset Splits | No | The paper describes generating 'mapping scenarios for n agents and targets' and performing '10,000 games' or '2800 drop-in player matches' in simulation environments, but does not specify exact training/validation/test dataset splits. |
| Hardware Specification | Yes | Table 2: Average running time (ms) of algorithms for values of n on an Intel(R) Xeon(R) CPU E31270 @ 3.40GHz. |
| Software Dependencies | No | The paper mentions software like 'Sim Spark', 'Open Dynamics Engine (ODE) library', and 'Agent2D base code release' and 'C++ implementations', but does not provide specific version numbers for these software components or libraries. |
| Experiment Setup | Yes | In UT Austin Villa s positioning system players positions are determined in three steps. First, a full team formation is computed using Delaunay triangulation (Akiyama and Noda 2008) based on set offset positions from the ball (formations used are provided in (Mac Alpine et al. 2013)). Second, each player computes an assignment of players to positions in this formation according to its own view of the world using the MMD+MSD2 role assignment function. An important factor in any SCRAM-based system is that agents have reasonably accurate knowledge of where all agents are currently located. We use agent communication to share and synchronize agent world models as discussed in (Mac Alpine, Barrera, and Stone 2013). For the third and final step a voting coordination mechanism detailed in (Mac Alpine, Barrera, and Stone 2013) synchronizes players computed assignments. |