Dyson Project N223

[assuncap]

Looks like Dyson's bot is finally ready to see the light of day.

16 years. 200 engineers. £28m investment.

So I guess It better be d*** good. Can't wait for September 4th

Edit: Summary and link to this discussion has been promoted to the front page:

http://www.robotreviews.com/news/2014/8 ... bot-vacuum

[Fraggboy]

If this is what Dyson wanted, better watch out iRobot, Neato, Samsung, and Karcher!!!!

[assuncap]

If this is what Dyson wanted, better watch out iRobot, Neato, Samsung, and Karcher!!!!

As long as it is priced correctly. Dyson products have always a premium price, 500-600 Euro is the limit for me.

[vic7767]

Boy, what a teaser.......This really could be something new and interesting.

[Fraggboy]

Boy, what a teaser.......This really could be something new and interesting.

Right?!?!

[robocleaner]

Looks like Dyson's bot is finally ready to see the light of day.
http://youtu.be/L6ReNFlxqJc

16 years. 200 engineers. £28m investment.

So I guess It better be d*** good. Can't wait for September 4Th

To take 16 years, 200 engineers and £28m might just mean they're crap at robotics and/or hugely inefficient... and it's taken this long, that many people, and that much money to understand concepts which the other makers sussed in less time, with fewer people, and less money!

[mfortuna]

You have to remember this is marketing. Those numbers could be derived from anything. That may be Dyson's total engineering headcount and budget for 16 years, not just for developing a robotic vacuum.

IMHO, Dyson's product line is all about marketing. I don't own any Dyson products so I can't comment on their functionality. But slick designs and slick advertising doesn't mean the end result will be a great product.

[Fraggboy]

We own the DC65 Animal, and absolutely love it!

You could be right.. Just nothing but Hype! BUT, if Dyson is true with what he has said in the past, than this robotic vacuum will be number 1.

Time will tell in the real world.

[robotreviews]

third_deg posted a link to this patent on June 2nd. The patent was filed on January 13th, but wasn't published publicly until January 16th, 2014.

viewtopic.php?f=6&t=17877

It has quite a few illustrations as well as a fairly detailed description of how it works. Here is the abstract:

"An autonomous vacuum cleaner comprising a chassis, a traction part for supporting the vacuum cleaner on a surface, a driving part for driving the traction part and a control system configured to control the driving part to guide the vacuum cleaner across a surface to be cleaned. The vacuum cleaner further includes a cleaner head having a dirty air inlet facing the surface to be cleaned and a separating apparatus carried by the chassis and communicating with the cleaner head in order to separate debris from an airflow entering the separating apparatus via the dirty air inlet. The separating apparatus comprises a first upstream cyclone and a plurality of second cyclones arranged in parallel with one another and located downstream of the first cyclone."

[third_deg]

It will map via computer vision, it will have full suction power. It will also be very large and give up some important elements to the task of robotic vacuuming in search of its primary goal of power and navigation.

[Fraggboy]

Hmm.... I'm intrigued.

[mofan]

Its not a 100% percent certainty, but it looks almost certain that Dyson is set to announce a robot vacuum on Sept. 4. That would be huge as it would pose an immediate and serious threat to Irobot. Think of all the features that you wish Roomba had, because they are now going to become reality much sooner than expected. Up until now, Irobot has had no viable market competition and has been free to innovate at its own (somewhat slow) pace. Dyson already is viewed as the world leader in quality and innovative vacuums, so if they continue that trend and can hold to a sub-$1000 price, Roomba will need to quickly improve to keep up: http://gizmodo.com/dyson-teases-upcomin ... 1627892075

[mfortuna]

FYI I merged mofan's separate Dyson topic into this one.

[mofan]

Sorry guys. Didn't notice this other thread. Really looking forward to Sept. 4.

[assuncap]

I had a look on both Dyson's Youtube and Twitter accounts and they share the same background image.

It looks like a camera sensor with a omnidirectional lens, exactly as the video footage would suggest.
The robot will have to have some serious processing power for this to work... or they do there processing elsewhere. Maybe using the base (or even the cloud) to do all the heavy processing and therefor freeing battery power for the vacuum. That would be very cool, engineering wise , but very unlikely.

[Fraggboy]

In the video it seems there are 4 cameras so it can have a 360 degree view.

[third_deg]

Continual image comparisons allow for range estimation, object recognition, and location marking. It learns the environment basically.

Dyson has been leaking bits of info over the years, and his lead scientist is best known for this work.

His patents are not a steep departure from his 2004 effort. Clearly better of course, more compact with better nav, but functionally very similar

The track based mobility was interesting. I'm very curious to see if that makes the final cut.

[assuncap]

In the video it seems there are 4 cameras so it can have a 360 degree view.

having 4 cams does't make sense, they can have the same results with 1 cam and a omin dir lens. Less parts, simpler design, no moving parts unlike the neato Ladar, the software does the magic.
The really tricky part is how to figure out depth with only 1 camera (no IR point cloud projections like the kinect). I read something, a few years back, about an algorithm developed by a English university professor who's working with Dyson on the whole robotic vision research. He seems to have cracked the depth problem.

[assuncap]

The track based mobility was interesting. I'm very curious to see if that makes the final cut.

That would solve the whole depth problem. If your able to calculate how many pixels a particular object has moved after turning or moving the robot, they would be able to triangulate the distance to the object. For that they need an IMU on the robot, low cost IMUs are on every smartphone now a days.

[mofan]

Irobot stock price has taken a hit yesterday and today, so it seems investors are a bit nervous about Dyson. There is one bit of hope for Irobot, though. Dyson makes great products, but they tend to be very expensive. Undoubtedly Dyson will have a cooler robot than the Roomba, but I am not so sure it will have a mass market price. Dyson is not going to displace the Roomba if it charges $1500 for its product.