Wednesday, April 1, 2009

Honda: Our in-house research can't read

Apparently, Honda has decided that they are the first to use a BCI (NIRS and EEG combo in this case) to control a robot (video here-and, oh, is it worth it). Not the first to use this technique, but the first to do it, ever. Okay, let's pretend that the Fetz 1969 experiments drove a 'device' instead of a robot (I'm not even sure how far back a device driven by surface electrode input might have been), but this is massively misleading. Sure this is just a stupid company statement, but Google "first robot controlled by a BCI" in two days and this crap will be the #1 result. *facepalm* I'm sure I can't really have an impact on such a large corporation's ridiculous press statement, so instead enjoy this video of ASIMO falling down stairs:

Thanks Katie for the heads up.


Brendan said...

Mr. King is correct in noting that virtually every jackass who has a robotic device driven remotely by brain activity claims to be the first. The lack of journalistic integrity and oversight is really harrowing.

For the record, the first reputable report of a mechanical device controlled by a BCI stems from Grey Walter in 1964. He used an invasive BCI to control a slide projector. He gave a talk about this to the Ostler Society in 1964, but never published it. Andy Schwartz and I both tried to find a publication about this, but could not. It is referenced in Consciousness Explained by Daniel Dennett.

I also asked my boss, Gert Pfurtscheller, who used to work for Gray Walter. He confirmed it. So, despite the lack of a peer reviewed pub, I consider Grey Walter's work valid, and the first BCI.

Dr. Brendan Allison
TU Graz

Brandon King said...

I had completely forgotten about Walter, good call.

ganges said...

Dear Brendan & Brandon, I loved the blog and came to similar opinion!

However, I think a technology assuring 90% (on single trials?) with 4 class problem is worth giving it some attention. Such a huge systems could be used for the rehab purpose than as a take-home device! What do you say?

BCI team @ EPFL

Brandon King said...

Meh. That's what I say. If it takes a minute to get that 90% classification with intense concentration, there are much more efficient ways of getting control signals (EMG, EOG, hell, EEG should be better than that alone). I think the consensus is that Honda was way out of line with the press BS.

Post a Comment