Subscribe to RSS
Archive for the ‘telepresence’ Category:
Whitepaper provides unconventional perspective on “videoconferencing”
I have released a brief whitepaper summarizing my contrary position with respect to “videoconferencing”. The paper is titled: Practical Applications of Low-cost Network-Based Video: Beyond Videoconferencing as a Substitute for Face-To-Face and is available at the following link: http://www.toyz.org/whitepapers/video_social2009.html
This paper was originally written in 2003 but I was not able to publish until now, for various reasons. But I stand by the research and conclusions.
Some of the more controversial positions include:
The present work totally abandons video as a substitute for face-to-face communication, and in fact suggests uses that are not videoconferencing, in any existing sense, at all. Further, we conclude that another common application of videoconferencing, that implemented by most desktop software, specifically the person-to-person video call, is nearly void of utility, particularly for business communications.
Please look it over and flame away.
More unconventional video wisdom
Tsahi Levent-Levi at the TMC Talking Video blog refers to my recent post on video in which he says:
Andrew MacDonald tried to see what’s his perceived audio quality threshold. He shows that improving audio quality improves the medium and the cues it provides.
Ok, no problem. Many years of research bear this out. Audio quality has a direct impact on the effectiveness of interactions. But then Tsahi takes a leap:
Video adds visual cues which are not available in voice calls. And high definition video gives more visual cues than its SD counterpart.
This is where the research does not support Tsahi’s assertion. Video can add visual cues, but not the ones people expect and assume. Further, audio provides more non-verbal cues than is usually assumed.
Ten years ago, Steve Whittaker at ATT research looked at research going back many years prior to that, and found:
We first evaluate evidence for the utility of providing non-verbal communication information using video. We conclude that previous work has overestimated the importance of supplying non-verbal information at the expense of speech.
And nothing has changed today. There is still way too much assumed about the value of non-verbal communication potential in video. People put way too much focus on this and, therefore, neglect other, much more effective, uses of video, such as for presence, shared context, and its role in facilitating opportunistic interactions.
Furthermore, while we often overestimate the value of video providing non-verbal cues, we also tend to under-estimate the degree to which audio provides non-verbal cues. Other research cited in the above paper has shown this to be true as well:
The research (Chapanis) compared two media conditions: audio only communication, and high quality video/audio. If video does indeed provide useful cognitive cues, then there should be benefits for providing visual information in these types of collaborative problem solving tasks, where it is important to track the understanding and attention of remote participants. However, the studies showed that adding visual information did not increase the efficiency of problem solving, or produce higher quality problem solving.
We humans tend to assume that video must provided these non-verbal cues. It’s so intuitive and obvious. But when one looks at it objectively, again and again, the data proves otherwise.
It’s easy to say video does provide such cues, casually. No one would question it, unless they’ve looked at the years and years of research.
We should note that Mr. Levent-Levi works at RADVISION, a company that makes video-conferencing products. They need to go where the money is – and that’s in people beating the 20-year long dead horse of “talking heads” video as a substitute for face to face meetings – not in far more effective, but less sexy, uses of video.
iSpykee – Open-source Spykee for iPhone
I finally finished making my previously mentioned iPhone hack for Spykee into a form I could release for public consumption. It’s still not perfect, but at least I’ve removed external dependencies so it can be installed without too much difficulty.
Spykee wifi robot interface for iPhone and iPod Touch
Basically you run the “controller” software on your LAN (the same LAN as the Spykee robot) and then use your iPhone to control and interact with your Spykee robot from anywhere.
The “controller” software is available as open-source and can be used on Mac OS X, Windows XP/Vista, or Linux/UNIX systems. As far as I know, this is the first release of an open-source implementation of the (binary) Spykee protocol. This software is provided in “C” under the BSD license, so it could be used as a basis for other home-grown Spykee applications, including motion detection, stealthy audio snooping etc.
iSpykee currently supports moving the robot, by touching areas of the video image: left, right, forward, and back; changing the robot motor speed (”Turbo mode”); turning the headlight on and off; taking a snapshot of what Spykee is seeing; turning “Video surveillance” (motion detector) mode on and off. It also supports a “low bandwidth” mode that can be useful when using iSpykee from a slow network connection (such as Edge).
Please check it out and join the iSpykee Google Group.
Updated to note that the “controller” software now works on Windows too.
Update on my eComm presentation
This has not made it to the eComm site yet, but here’s what my 5-minute lightning talk will be about:
Mobile phones are practically ubiquitous. Everyone carries them and most of them are turned on and connected 24×7. Today’s mobile phones, even the least expensive ones, have sensors, spare cycles, and connectivity. These resources can be applied to a wide variety of Social Telemetry Applications, to powerful, and potentially even troubling, effect. We present findings of a real-world deployment of such a system.
And here’s a sneak-peak of the talk:
eComm presentation Wednesday, March 4, 4:55PM Salon E
I look forward to seeing you there.
Not so startling Cisco research on video
A tweet from Dina Mehta via Mark Petrovic pointed me to an article titled The psychology of videoconferencing that refers to some research Cisco recently released. Cisco says:
[research published in 1971] revealed that only seven per cent of our understanding comes from pure words, and that 40 per cent is gleaned from the tone of the voice and 53 per cent from visual cues.
What they fail to note is that all the research has shown that video is not very good at providing those “visual cues” especially more subtle ones.
For many years, videoconferencing technologies have been applied as a substitute for face-to-face meetings. Spending for videoconferencing technology is typically based on a presumed increased productivity or savings due to a presumed reduction in travel. Despite countless deployments and projects deployed over decades, on the whole, the technology has failed to provide these anticipated benefits.
When videoconferencing has failed to meet expectations, it has almost always been attributed to some “other” factors. There seems to be a powerful intuitive desire in people to find something else to blame. I was guilty of this myself, for a long time. We keep telling ourselves some detail of the implementation needs improvement. If we just used a better camera here, a little more bandwidth there, an improved user-interface, one more update of some kind, all will be well.
However, the answer is right there in prior research. We will find that many others before us had fallen into the same trap. And it’s not because these projects were using old hardware. There have been some deployments using absurdly expensive hardware and software that would never be practical in a wide deployment. And even such sophisticated systems could not deliver on the presumed power of videoconferencing technologies to substitute for face-to-face communications. It wasn’t the hardware or network. The most expensive hardware in the world can’t address some of the fundamental limitations of real-time videoconferencing.
The elephant in the room is that even high quality audio and video cannot replicate the rich nature of face-to-face communication. Period.
Further, experience shows that this is very very difficult for people to accept. It is much easier to find something else to blame. We tend to point at problems in the details of the specific implementation rather than accept the reality that real-time videoconferencing is inherently limiting and then work within those limitations.
In Cisco’s paper, they say:
“We observed the value of visual cues in successful meetings, and video technologies that maximise this, such as telepresence, are ideal for maintaining excellent relationships.
“However, individuals who approach meetings with a positive attitude, leaders who understand and support the different personalities and cultures in their teams, and organisations that provide the resources and training to make video communications the norm, are also essential to effective video-enabled meetings.”
So what they’re saying is when the technology doesn’t work, or doesn’t give all the benefits people expect (because it never does), it’s your fault because you failed to “understand and support the different personalities and cultures in your teams.” Nice preemptive scapegoat, Cisco.
Spykee iPhone Hack
At left you’ll see my first cut at a rudimentary Spykee mini-console for iPhone. If you’re familiar with the standard PC or Mac Spykee console app, you’ll recognize many of the functions.
This app only implements a sub-set of the full PC and Mac consoles. I basically wanted to get some minimal functionality working before going too crazy with the UI and more esoteric features. Here’s what it will do:
(1) Move the robot, using the left, right, forward, and back buttons. Each press of the button moves a fixed amount.
(2) Turn the camera LED light on and off.
(3) Tell Spykee to take a snapshot.
(4) Turn ‘Video surveillance’ mode on and off.
Most importantly, the app shows you on the iPhone what the Spykee is seeing. It does not stream video in this version. It updates the “viewport” image periodically when the scene changes (or on demand with the ‘Update’ button). This keeps bandwidth usage down and still gives you a view of what the robot is seeing — and it even works on slow Edge connections.
It’s got another feature to help when controlling Spykee from a slow iPhone connection, where using the ‘Resolution’ button, you can flip between a lower resolution (but much less data) image and the normal full resolution Spykee view. This, and other features of the app, are demonstrated in the video below.
I made a video to show that, while this is still a hack, and is not ready to distribute by any means, it does actually work in real life, with a live robot. If I just provided the above screenshot, there’s no way to tell if it’s just a mock-up.
The video shows some of the basic functions as well as limitations. Even with those limitations, it is pretty cool (if I do say so myself), to be able to sign in and view what Spykee is seeing from anywhere, with just the iPhone without lugging around a PC or Mac, and even on a slow Edge connection.
UPDATE: Finally released. See: http://mrblog.org/2009/03/15/ispykee-open-source-spykee-for-iphone/
Spykee wi-fi robot – first looks
I saw the Spykee robot at Costco and so I went home and researched what it was all about.
Then I decided to run back to Costco and pick one up before the US supply runs out. It was $229.99
Spykee was created by a French company, Meccano and is being distributed in the USA under the Erector brand. The robot has been available in the UK and Europe for some time – it was at one time called Spyke, but probably as a result of some trademark issues, now is called Spykee.
The basic idea is that the robot connects via wifi and therefore can be controlled either locally on the same wi-fi network, or remotely from any IP address. It’s sort of a remote-controlled webcam that you can drive around.
It’s supposedly a kid’s toy, but I’m not sure it’s going to be all that fun for kids, frankly. It’s very cool, and I hope to do some fun (and perhaps even interesting) stuff with it, but here are a few of my gripes:
- The building process is overall poor. The instructions are terrible. The fasteners that Meccano provide with it are really junk and awful for load bearing – I substituted my own nylon screws with nuts to end the thing falling apart at the slightest touch/bump.
- Meccano support sucks. Well actually, it just doesn’t exist at all. The toy doesn’t come with a CD or printed manuals. You have to download them. But the catch is that Meccano doesn’t seem to know how to operate either a web server or a mail server. Their web server takes users to the UK site with no links back to the US site, for reference, you can get to the US site by manually entering this address in your browser: http://www.spykeeworld.com/US/
- As I note, they can’t run a mail server either, apparently, because mail to the address they list for support bounces (and it has been this way for a long time, according to forums around the net).
- The robot looks like it has arms that move, but the entire body of the robot is just a frame to hold the webcam. It serves no function and doesn’t do anything (cannot be moved via remote control). The “active” parts of the robot are the base with the tracks and CPU etc. and the webcam “module” which also houses the LED light and microphone. The rest of it is inanimate, just for show.
- The Spykee is not very autonomous. It only does things while one is connected to it from the “console” software. You can’t activate a function, then log out, and expect it to do anything (like act as a surveillance camera).
- There is no web interface – the only way to connect to the robot and interact with it is to use a specific binary application (Mac and Windows supported). So that app has to be installed on any computer you want to use to interact with Spykee and those computers have to be a Mac or Windows PC (i.e. no iPhone or othe such).
- The name has the word “Spy” in it, but Spykee is not very stealthy. It’s not going to sneak up on anybody (It’s LOUD).
- You can theoretically talk beween the remote PC and somebody near the robot, but it doesn’t work very well, at least not on the Mac version. It has horrible delay and no echo handling, so is near useless. If you mute the mic on the PC side, you can use this feature to listen to sounds near the robot, but it’s hard to interact.
- The manual says the software is “open source” but it is nowhere to be found (people keep saying it will be released, but there’s no dates anywhere that I’ve found).
There’s a good YouTube video out there listing some of the above and a few other criticisms here.
I got Spykee because it supports the Mac. The other wifi robot, Rovio, only works with Windows.
I haven’t done anything with the Spykee yet except the “officially supported” things, which are actually pretty cool, but limiting. I want to connect to the device directly with my own software, with web services etc. Unfortunately, the protocols are not released, nor is the supposed “open source software”, so this will require hardcore reverse engineering. I haven’t spent any time on that yet, so I can’t provide any details yet.
In getting the remote control mode to work, you setup a name/password for your robot on SpykeeWorld.com and then connect from a remote place using that name. One thing I found out is that this name/password must be simple letters with no spaces or other punctuation. It will let you set a name with these characters, but when you try to connect, it won’t work (and the diagnostic “recipient not available” is not helpful). I have been able to use the remote access (from outside on the Internet to the robot behind a NAT/firewall) with the robot sitting behind many different makes and models of firewalls and routers, and I even tried two layers of firewalls and that worked too, which surprised me. I’m not sure yet how it gets through the NAT/firewall.
Like I said, I rushed out and bought it because I didn’t want to miss this wave of US shipments, but before you do the same thing, you might want to be aware of the above caveats.
Videoconferencing predictions past and present
This week we’ve seen many references to “recently released” research suggesting that videoconferencing may finally be ready to take off.
Videoconferencing – a much talked about, but seldom-used technology – may finally be earning its spurs within the enterprise.
85 per cent of respondents polled either use or plan to use video conferencing.
Forty one per cent of those questioned are also using, or investigating the use of so-called telepresence
But here’s the thing: We’ve seen this before. For example:
“Videoconferencing gains momentum”
Jul 18, 2007“Video Conferencing Gains Ground”
Sep 17, 2007“Videoconferencing is likely to gain momentum”
Apr 4, 2007“Videoconferencing Gains Momentum”
Sep 7, 2001
Despite years of false starts and unfulfilled promises, here we go again. Whether we are talking one year ago, seven years ago, or 25 years ago, the theme is always the same. It goes like this: “while we acknowledge that videoconferencing has never met adoption expectations in the past, we believe that is because of INSERT EXCUSE HERE and with our new INSERT WIZBANG TECHNOLOGY DIFFERENCE HERE we are sure adoption will explode this time.”
Hint – it’s not a technology problem.
FYI: Telepresence is not “new”
Cisco is pushing Telepresence hard as “the next big thing” and a lot of people are buying into the hype and referring to it in language like “new” and “next generation” technology.
I have news for you. Telepresence has been around for decades, at least in research circles. It used to be called Videoconferencing and it is usually justified on the basis of increased productivity or cost-cutting as a direct result of a presumed reduction in travel. Except it turns out it doesn’t achieve those benefits.
And it’s not because the implementations haven’t been good. There have been some magnificent systems put together (at incredible cost), with high quality video, multiple cameras, synchronized whiteboards, and all sorts of cool hardware and features.
We see solutions claiming to provide technology that substitutes for face-to-face communication come up over and over. However, no one seems to ever stop and look at the facts, the research already done, the long history of failure in past attempts, etc. It seems we are doomed to repeat this history forever. Its too compelling – intuitively we know it must work.
But it doesn’t. I mean the technology works wonderfully – but it doesn’t achieve the stated goals of increasing productivity or reducing travel expenses. I’ll make this part one and end here, continuing this theme tomorrow.
