eBay Auctions for the Robopet are expiring now!

eBay Auctions for the i-Cybie are expiring now!

Friday, December 09, 2005


Robopet Roll-Over Quirks

Some owners of robot dogs have reported a strange quirk when their Robopet does theRoll Over trick. Normally the robot flips over on it's back, waggles a bit, and then completes a full roll by flipping upright in the same direction that it flipped on its back. My chrome Sharper Image Robopet does exactly that. But some people have reported that their doggy flips upright in the opposite direction. So instead of doing a full roll, it does a half roll and then unrolls itself back to the same position it was in before it began the trick.

I am not sure what causes this. The Robopet uses tilt sensors to detect when it has fallen, and then uses internal algorithms to decide how to right itself. This is purely a guess, but it is possible that the Robopet's electronics, due to variances in the tilt sensors between models and variances in the power and sway in the waist motor, thinks that the best way to get back up is to go back the way it came, instead of completing its roll.

If anyone has their own story to tell about their Robopet doing the Roll Over trick, or if someone knows the real answer to this puzzle, please leave a comment.

Very interesting. Even though my wife and I have spent hours playing with Robopet, we never paid attention to which direction he was getting up.

A look through several of the Robopet videos on Robot's Dreams (http://www.robots-dreams.com) provides some data points-

In this video - http://www.robots-dreams.com/2005/11/teaching_a_new_.html - you can see that our Robopet tends to get up on the same side.

In this video by Matthais Matting - http://video.google.com/videoplay?docid=-6533640395361374223&q=robopet - you can see that his Robopet doesn't go down hard - it seems to just go down part of the way, then comes back up. Later in the video he manually turns the Robopet on its back, and it gets up to the opposite side.

It would take some additional testing, but my guess is that the Robopet doesn't really "roll-over". Instead, it "falls-down" and "gets up". If it falls down with enough momentum so that it comes to rest lying on its opposite side, then it gets up in that direction.

Things like the battery state (fresh or weak), battery weight (more or less mass and momentum), the surface (slip or get traction when starting the fall move) might contribute to the variation in performance as well.

Thank you for that detailed report! It does appear that the waist motor in the Robopet is a big variable between units. Strange.
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?