Group Actions
Engagement
  • Author Author: easyejl
  • Date Created: Date Created
  • Views 6556 views
  • Likes 9 likes
  • Comments 52 comments
Related
Recommended

Hey Moogle - an animatronic moose assistant

easyejl

My wife and I went to see Evil Dead - The Musical, and I was entranced by the moose puppet on the wall. I thought "Heck, it would be easy enough to make an animatronics moose to have at home", so I set out on this project.

 

 

Basic plan steps are

  • Acquire moose head puppet
  • Acquire robotic arm + servos
  • Begin programming pi - Basically can use most of the Google AYO voice pieces, need to add some references to pulse audio to capture the sound as he's speaking, to use to move the mouth
  • Assemble pieces
  • Laugh a lot

 

Steps completed

Moose head puppet on the way from ebay!

image

Robotic arm on order

Google AYO kit reformatted back to original and programming has begun!

Parents

  • High in the list of things that make me want to take a sledgehammer to a raspberry pi...

     

    • I set up a fresh install.
    • Got the deprecated google library (vs sdk) working with a USB mic + headphone out to powered speakers
    • Partially due to advice here from another thread I decided to do incremental backups
    • Setup samba
    • Setup a share on my pc
    • Used rsync to copy all files to a folder on the pc
    • 7zipped the folder with a relevant name

     

    So this way I know I can restart from that working. Then I installed pulseaudo, verified everything was working and backed up again, zipping to a new file after it was done.

     

    And then I installed the Waveshare WM8960 sound hat which is where the problem has begun before. The specific reason for that hat is that it has built in microphones (stereo) and a small built in amplifier plus it came with speakers for $20. that way it stays compact and doesn't need yet another power source in the moose.

     

    Sooooo any guesses what happened? after a few pokes at the asound config, the damn thing is working fine, 100%. image so its running the google assistant via hotword just fine. That's using the demo only program which is ok, but now I can at least clone the google samples that match that library version and hack in another voice for it as output. I may leave it as is, because looking at the raw data, not every stream back comes with text data. I'm still not sure why that is, when I ask what time it is, the assistant just answers - but the answer packet doesn't include it as text for some reason.

     

    anyhow, this means that with this backed up, I can concentrate on the sound recording/monitoring part, to get the audio output volume / frequency and then run a servo based on that. Should be relatively easy, and now I can revert to this working on the hardware I wanted at any time even if I have to re-install

Comment

  • High in the list of things that make me want to take a sledgehammer to a raspberry pi...

     

    • I set up a fresh install.
    • Got the deprecated google library (vs sdk) working with a USB mic + headphone out to powered speakers
    • Partially due to advice here from another thread I decided to do incremental backups
    • Setup samba
    • Setup a share on my pc
    • Used rsync to copy all files to a folder on the pc
    • 7zipped the folder with a relevant name

     

    So this way I know I can restart from that working. Then I installed pulseaudo, verified everything was working and backed up again, zipping to a new file after it was done.

     

    And then I installed the Waveshare WM8960 sound hat which is where the problem has begun before. The specific reason for that hat is that it has built in microphones (stereo) and a small built in amplifier plus it came with speakers for $20. that way it stays compact and doesn't need yet another power source in the moose.

     

    Sooooo any guesses what happened? after a few pokes at the asound config, the damn thing is working fine, 100%. image so its running the google assistant via hotword just fine. That's using the demo only program which is ok, but now I can at least clone the google samples that match that library version and hack in another voice for it as output. I may leave it as is, because looking at the raw data, not every stream back comes with text data. I'm still not sure why that is, when I ask what time it is, the assistant just answers - but the answer packet doesn't include it as text for some reason.

     

    anyhow, this means that with this backed up, I can concentrate on the sound recording/monitoring part, to get the audio output volume / frequency and then run a servo based on that. Should be relatively easy, and now I can revert to this working on the hardware I wanted at any time even if I have to re-install

Children
No Data