A Go implementation of an optimal Yahtzee player on the Raspberry Pi. See more about this project on my blog, or play along online.
You will need to build the score tables first, using the compute_scores tool:
$ go get github.com/timpalpant/yahtzee
$ cd cmd/compute_scores
$ go build
Build the expected value score tables:
$ ./compute_scores -logtostderr -observable expected_value -output expected-value.gob.gz
Build the high score tables:
$ ./compute_scores -logtostderr -observable score_distribution -output score-distribution.gob.gz
The expected value tables are 5.7 MB and the high score tables are 1.8 GB on disk.
Build and run the web server, passing the location of the score data tables:
$ go install github.com/timpalpant/yahtzee/cmd/yahtzee_server
$ yahtzee_server -logtostderr -port 8080 -expected_scores expected-scores.gob.gz -score_distributions score-distributions.gob.gz
The server will take a few minutes (and GB of RAM) to load the score tables at startup. Navigate to http://localhost:8080.
The image processing web service is written in Python. The easiest way to build and run it is with Docker:
$ cd image_processor
$ docker build -t yahtzee-image-processor:latest .
$ docker run -p 8080:8080 yahtzee-image-processor:latest
A pre-built image is also available at: docker.io/timpalpant/yahtzee-image-processor:latest.
First install GoCV, as described here: https://gocv.io/getting-started/
Then build the RPi player with:
$ cd cmd/rpi
$ go build
and run with:
$ ./rpi -logtostderr -v 3 -yahtzee_uri http://localhost:8085 -image_processing_uri http://localhost.local:8080 -score_to_beat 300
It is recommended to run the yahtzee server and image processing service on a separate machine, since they require more memory and computational resources than available on the pi.
This package is released under the GNU Lesser General Public License, Version 3.0
Yahtzee is a registered trademark owned by Hasbro. Hasbro is not affiliated with this project.