Explorer Box AR+

Augmented Reality Sandbox - simply mold the sand by hand and the landscape comes to life

Hardware

You need following hardware:

• Raspberry Pi 4 - 2GB RAM
• Microsoft Kinect Sensor v1 (XBox 360) + Power supply (USB adapter and charger cable)
• Beamer BenQ MX631ST (Short-throw, ratio 4:3)
• Assembled sandbox (e.g. wood)

Instead of Raspberry Pi you can use any other computer e.g. Intel NUC running on Debian

Installation

Install on Raspberry Pi latest version of Rasbian (desktop version). You can find rasbian images here

Update sources

sudo apt-get update
sudo apt-get upgrade

Install prerequisites

sudo apt-get install cmake libudev0 libudev-dev freeglut3 freeglut3-dev libxmu6 libxmu-dev libxi6 libxi-dev git python

Download and build libusb

Next we will build libusb, which is required by libfreenect. You can find the libusb releases here

mkdir ~/src
cd ~/src
wget https://github.com/libusb/libusb/releases/download/v1.0.23/libusb-1.0.23.tar.bz2
tar -jxf libusb-1.0.23.tar.bz2
cd ~/src/libusb-1.0.23
./configure
make
sudo make install

Download and build libfreenect

Finally download and build libfreenect. You can check for latest version at the libfreenect Github page.

cd ~/src
wget https://github.com/OpenKinect/libfreenect/archive/v0.5.7.tar.gz
tar -xvzf v0.5.7.tar.gz
cd ~/src/libfreenect-0.5.7
mkdir build
cd build
cmake -L ..
make
sudo make install

Clone repo

git clone https://github.com/mbit-solutions/explorer-box

Run

Kinect mode

Kinect sensor must be connect to Raspberry Pi

sudo python main.py

Fake mode

No kinect sensor necessary - generates random depth image

sudo python main.py fakenect

Calibrate kinect

Kinect sensor must be connect to Raspberry Pi

sudo python main.py kinect_calibrate

Calibrate beamer

sudo python main.py beamer_calibrate