Multi-Agent Pathfinding with Continuous Time
Authors: Anton Andreychuk, Konstantin Yakovlev, Dor Atzmon, Roni Stern
IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conducted experiments on grids, where agents can move from the center of one grid cell to the center of another grid cell. The size of every cell is 1 1, and the shape of every agent is an open disk whose radius equals 2/4. This specific value was chosen to allow comparison with CBS, since it is the maximal radius that allows agents to safely perform moves in which agents follow each other. To allow non-unit edge costs, we allowed agents to move in a single move action to every cell located in their 2k neigh-borhood, where k is a parameter [Rivera et al., 2017]. Moving from one cell to the other is only allowed if the agent can move safely to the target cell without colliding with other agents or obstacles, where the geometry of the agents and obstacles are considered. The cost of a move corresponds to the Euclidean distance between the grid cells centers. |
| Researcher Affiliation | Academia | 1Federal Research Center Computer Science and Control of Russian Academy of Sciences 2National Research University Higher School of Economics 3Peoples Friendship University of Russia (RUDN University) 4Ben-Gurion University of the Negev |
| Pseudocode | No | The paper describes algorithms textually but does not contain a formal pseudocode block or algorithm figure. |
| Open Source Code | No | The paper mentions 'We thank the EICTS authors who made their implementation publicly available (https://github.com/nathansttt/hog2).' but does not state that the code for CCBS is open-source. |
| Open Datasets | Yes | Next, we experimented with a larger grid, taken from the Dragon Age: Origin (DAO) game and made available in the movingai repository [Sturtevant, 2012]. |
| Dataset Splits | No | The paper states 'For every number of agents we created 250 different problems.' and 'Start and goal states were chosen randomly, and we create 250 problems for every number of agents.' It does not provide explicit training, validation, or testing dataset splits. |
| Hardware Specification | No | The paper does not specify any hardware details such as CPU, GPU models, or memory used for the experiments. |
| Software Dependencies | No | The paper mentions 'OMPL library' and the 'E-ICTS authors who made their implementation publicly available' but does not provide specific version numbers for any software dependencies used in their own experiments. |
| Experiment Setup | Yes | We conducted experiments on grids, where agents can move from the center of one grid cell to the center of another grid cell. The size of every cell is 1 1, and the shape of every agent is an open disk whose radius equals 2/4. This specific value was chosen to allow comparison with CBS, since it is the maximal radius that allows agents to safely perform moves in which agents follow each other. To allow non-unit edge costs, we allowed agents to move in a single move action to every cell located in their 2k neigh-borhood, where k is a parameter [Rivera et al., 2017]. |