I did that a few years ago, although it used a Kinect: https://pdfs.semanticscholar.org/b07f/4732e5d631cf182a59b1f28a9b458ddda785.pdf https://pdfs.semanticschol

I did that a few years ago, although it used a Kinect: https://pdfs.semanticscholar.org/b07f/4732e5d631cf182a59b1f28a9b458ddda785.pdf https://pdfs.semanticscholar.org/b07f/4732e5d631cf182a59b1f2... . It works with gestures, without a projected keyboard and has some audio and visual feedback on screen.