Boston Dynamics’ Latest Quadrobot using Velodyne LiDAR

Boston Dynamics have shown off their first robot since being acquired by Google, and it has one new component that’s very interesting.

On top of its head is a Velodyne HDL-32 LiDAR. They appear to be using it for obstacle mapping, that would then feed into their pathfinding algorithms. Those algorithms are interesting in their own right – a legged robot like this can avoid obstacles of certain sizes by simply stepping over them. The solution space is therefore pretty difficult to distill down into 2 dimensions.

I would love to know more about all of this, but there doesn’t appear to have been a press release and Boston Dynamics are typically pretty secretive given their clients include the US Army.

These guys are doing some very impressive work at the moment and it’ll be interesting to see what technologies come about using these robots as a starting point.

Everything is Complicated

It often amazes me how some things which appear trivial on the surface can actually lead to deep optimization problems. Take for example the programming challenge, Elevator Saga. Just make the lifts take the little people where they want to go. Seems easy, right?

It is – on the surface, and for the first few levels. But when you start getting more traffic and multiple lifts, things get very complicated. There even comes a point where lifts need to know the state of other lifts. Do you have lifts run in “express” mode, skipping some of the lower floors? When should they make the decision to return to ground? Do you pass by a waiting passenger if you’re ascending / descending at full speed?


JS1k Entry

I stumbled across a JS1k entry I made in 2010. It’s not much, but it has pretty colours. The competition aims to fit as much functionality as possible into 1 kilobyte of javascript code. I wrote a plasma zoomer with colours in 788 bytes (when minified).

Your browser does not support the canvas tag. This is a static example of what would be seen.