In map-predictive exploration planning, the aim is to exploit a-priori map information to improve planning for exploration in otherwise unknown environments. The use of map predictions in exploration planning leads to exacerbated greediness, as map predictions allow the planner to defer exploring parts of the environment that have low value, e.g., unfinished corners. This behavior is undesirable, as it leaves holes in the explored space by design. To this end, we propose a scoring function based on inverse covisibility that rewards visiting these low-value parts, resulting in a more cohesive exploration process, and preventing excessive greediness in a map-predictive setting. We examine the behavior of a non-greedy map-predictive planner in a bare-bones simulator, and answer two principal questions: a) how far beyond explored space should a map predictor predict to aid exploration, i.e., is more better; and b) does shortest-path search as the basis for planning, a popular choice, cause greediness. Finally, we show that by thresholding covisibility, the user can trade-off greediness for improved early exploration performance.
The full paper is available at
https://lericson.se/papers/ecmr21_understanding_greediness_in_exploration.pdf
Clone this repository somewhere.
git clone https://git.lericson.se/fragmenot.git/
Change into that directory.
Do this bit of black magic for dependencies:
git fetch origin +refs/heads/deps/*:refs/heads/deps/*
Create a virtualenv:
virtualenv --python=/usr/bin/python3 env
Install dependencies
Ubuntu:
apt-get install libspatialindex-dev liboctomap-dev libdynamicedt3d-dev build-essential python3-dev python3-numpy
./scripts/install-embree.sh
macOS:
brew install spatialindex octomap embree
[You need to manually install dynamicEDT3D?]
All:
./env/bin/pip install -r requirements.txt
./env/bin/python3 setup.py develop
Run the demo:
./env/bin/python3 src/__main__.py
The easiest way to run headless (i.e. without an actual display connected) is
to set the environment variable GUI_HEADLESS=true. This doesn't generate any
images though, so if you want that you will need the X virtual framebuffer,
Xvfb. Then
xvfb-run -s '-screen 0 1280x720x24' ./env/bin/python3 src/__main__.py
Most of what we do with this is to collect statistics. To make running batches easier, there is a system with the very funny name "batsh".
How to use: start a tmux, then the batsh-queue program.
./scripts/batsh-cmds | ./scripts/batsh-queue
You may run batsh-cmds to inspect which commands will be run, one per line of output.
Once the queue program is running, open new tmux tabs and start a worker in each:
./scripts/batsh-worker
This has the advantage that you can cycle through the current runs in tmux, with full terminal interactivity. Batsh is the future.
Use
./scripts/plot_runs.py ./var/runs/runs_foo/*