SMART: Self-Morphing Adaptive Replanning Tree

Zongyuan Shen, James P. Wilson, Shalabh Gupta*, Ryan Harvey

Department of Electrical & Computer Engineering, University of Connecticut, Storrs, CT, USA

[Paper] [Project Page] [Video] [Slide] [Citation]

Introduction
Usage
Citation
Acknowledgement
License
Maintaince

Introduction

SMART facilitates fast reactive replanning in dynamic environments. It performs risk-based local tree-pruning if the current path is obstructed by nearby moving obstacle(s), resulting in multiple disjoint subtrees. Then, it exploits these subtrees and performs informed tree-repair at hot-spots that lie at the intersection of subtrees to find a new path.

Fig. Point robot (yellow point) and dynamic obstacles (grey circle) move at a constant speed of 4m/s.

Usage

C++

User-defined inputs:
- File "Main.cpp": trialIndex, dynObsNum, dynObsSpeed, sceneIndex, dynObsPosition, robotInitState, goalState
- File "SMART.cpp": cellSize, staticObsMap.txt

trialIndex: seed for random sample generation
sceneIndex: seed for dynamic obstacle trajectory generation
dynObsPosition: initial position of dynamic obstacle
goalState: goal position
robotInitState: contain initial robot position and constant linear speed
cellSize: size of cell in meter
staticObsMap.txt: binary occupancy grid map. free = 0; occupied = 1.

Outputs:
- Replanning time (s)
- Trajectory length (m)
- Travel Time (s)
- Recorded data
Recorded data:
- Dynamic obstacle
- Tree
- Path
- Robot's footprint

Function dataRecord() is used to record the data for visualization on Matlab.

Compilation:

compile Main.cpp
run the executable file

Matlab for visualization:

User-defined input:
- "folder": It is a character array to store the directory to the folder that contains the recorded data.
- "videoRecord": A demo video will be created if videoRecord = true.
Demo generation:

run Main.m

Citation

If you use the results presented in this paper or the code from the repository, please cite this paper:

@article{shen2023smart,
title={SMART: Self-Morphing Adaptive Replanning Tree},
author={Shen, Zongyuan and Wilson, James P and Gupta, Shalabh and Harvey, Ryan},
journal={IEEE Robotics and Automation Letters},
year={Sep. 2023},
volume={8},
number={11},
pages={7312-7319}
}

Please kindly star ⭐ this project if it helps you.

Acknowledgement

This research is supported by the Air Force Research Laboratory.

License

Maintaince

For any technical issues, please contact Zongyuan Shen (zongyuan.shen@uconn.edu).

ZongyuanShen/SMART