This simulation is based Unity 3D game engine software. It simulates a multi-robot search and rescue (or foraging) problem with different robots having different knowledge base. For example some robots know only to pick up red/blue/green/yellow targets, while some robots know all targets or no targets. All robots have basic collision avoidance and random walk routine in their knowledge base. The simulator has a key option: "With KT-BT". When this option is enabled, the robots can communicate with each other and explicitly share their knowledge through Behavior Trees using the KT-BT framework.
If you use/refer to our work, please consider citing the below paper.
- S. S. O. Venkata, R. Parasuraman and R. Pidaparti, "KT-BT: A Framework for Knowledge Transfer Through Behavior Trees in Multirobot Systems," in IEEE Transactions on Robotics, vol. 39, no. 5, pp. 4114-4130, Oct. 2023, doi: 10.1109/TRO.2023.3290449.
The preprint of the publication is available in ArXiv: https://arxiv.org/abs/2209.02886
Click below for a video presentation KT-BT works with a demonstration of BT transfer.
Download the latest release here (v4.2).
After downloading the .zip file extract all the contents to a directory.
Open the file BTTransferSimulator.exe
You may have to click "More Info" and "Install Anyway" if your firewall blocks/scans the application.
Note: If you are using an antivirus or antimalware sofware, you may have to add BTTransferSimulator.exe and LearnBTV1.exe files to the exclusion list.
Once you run the application - the simulator settings Frontend will pop up. Choose your settings as described below and click "Run Simulations" button. It will take some time to initialize and run the Unity-based simulations (up to a minute in some cases at least in the fist trial) and eventually it will create a new window with the simulation visualization. Please wait for this process.
Target Type: There are four target types in the simulator that are colored red, green, yellow and blue and have four corresponding zones at each corner
Target Counts: The number of targets of each type to be initialized in the simulations.
Obstacles: Check to add static obstacles
Scenario: There are currently two scenarios available for the current study
- 'with KT-BT' and
- 'without KT-BT'
Robot Types: There are 3 types of robots,
- I -Ignorant - Robots that have no initial knowledge about any of the targets
- M - Multi-target - Robots that have knowledge about all the targets
- R/G/Y/B - Single-target - Robots that have knowledge about red, green, yellow and blue colors.
Communication Range: The range within which it can communicate with its neighbors. This is measured in times of the size of the robot.
Total Trials: The number of trials to conduct for data collection
Start Count: Starting trial number count for running simulations. This is to let the user continue simulations from the point they are stopped.
Continuous: If unchecked, simulations stop after each trial and wait for the user to press the exit button on the simulator. If checked a trial instance is closed and a new trial instance is opened automatically.
Parallel Instance Interval: This simulator supports parallel simulation instances. If set to a value more than 1 and if the trials are greater than 1 multiple simulataneous instances are initiated. For e.g. if the trials are 10 and the max parallel instance intervals is 3, only 3 instances are open at a time.
Note: Multiple parallel instances might slow down the simulations, hence a PC with good processing speed and a GPU is recommended
Duration: The maximum number of iterations before stopping. Each iteration has a fixed time of 0.01s.
Targets: Stop after a certain overall percentage of targets are collected.
No Stop: Simulations do not have stopping criterion.
Record Data: Enable this to record data
Auto Name: Auto Names the files and folders (Recommended)
Save Path: Choose save directory for the recorded data.
Config. File: Parameter summary
Simulation Control: Run Simulations -> Click to run simulations after setting the parameters.
Pause/ UnPause: Pause or unpause all instances of simulations
Stop Simulations: Stop all the instances/ terminates all simulation instances.
Note: Do not press stop simulations while the instances are initializing
The simulation parameters segment in an instant shows the parameters transfered from the settings form.
Camera View: Currently there are two views available.
# Robots that can handle: Counts of robots that have knowledge about each target type.
# Known Conditions: Number of robots that are ignorant (None), knowledge about 1, 2 3 and all targets respectively.
# Robot States: Number of robots in different states at a given point of time/ iteration.
-
Sanjay Sarma Oruganti Venkata -- PhD Candidate, College of Engineering, University of Georgia (https://github.com/sanjayovs)
-
Prof. Ramviyas Parasuraman -- HeRoLab, School of Computing, University of Georgia (https://hero.uga.edu)
-
Prof. Ramana Pidaparti -- DICE Lab, College of Engineering, University of Georgia (https://dice.engr.uga.edu/)
Heterogeneous Robotics Lab (HeRoLab), School of Computing, University of Georgia. http://hero.uga.edu
For further information, contact sanjaysarmaov@uga.edu or Dr. Ramviyas Parasuraman ramviyas@uga.edu
