Skip navigation
1 2 Previous Next


29 posts
The idea of smart Java dust becoming pervasive is compelling (consider the SunSPOTtechnology, including the Squawk JVM). If Sun really wants this to sweep the market, there has to be a path to make small commercial Java systems extremely easy to develop and affordable to license. Then, every individual or small shop with a bright idea can afford to implement it, and will want to. Yet the typical Java license fee is in the range of US$ six figures (per year). Does this license model support the kind of widespread adoption which Sun claims to want? "Wait a minute - isn't Squawk open source?" you ask. Yes, but 'open' is not necessarily 'free'. Squawk is free fornon-commercial use, while for-profit use requires the typical ME license fee mentioned above. Industry-wide, interest in 'smart dust' is becoming so widespread that other vendors (such as Texas Instruments, which now owns Chipcon, the vendor of the radio chips used in SPOTs) are launching low cost, open standards wireless sensors initiatives, based on the new generation of 2.4 GHz radios. The recent Embedded Systems Conference hosted a TI eZ430-RF2500seminar packed with 500 (my guesstimate) people who all walked away with an MSP430-based (the same chip family used in Sentilla motes) wireless sensor kit. The eZ430 kit (which is 2.4 GHz but not 802.15.4-based) is $49 including 2 nodes (one with a battery) and an Eclipse C-language toolkit. Or get the 802.15.4 (with free Z-stack Zigbee) versionfor $99. These prices are a long way from $10 (the oft-quoted price of a truly 'pervasive' dust mote), but not unreasonable. There are no hidden tools or license costs - TI just wants you to buy their chips. TI is not alone - you can find similar offerings from Cypress, Atmel and others. None of the TI solutions support Java (but the Sentilla option does - and look for some news from them at JavaOne). Sentilla mote prices are TBD and early dev kits are a few thousand dollars). The (more advanced) SunSPOT kits are now $750 to the general buyer, $300 or perhaps free to educational users. Yes, these are different devices -- SPOT is a much more capable platform in terms of processing capability and included sensors. So comparing the Sentilla or eZ430 to SPOT is not entirely fair. But if the goal is smart dust, these are the best offerings which each vendor has chosen to create. Someone else could create a lower cost SPOT and sell it... oops, well, they would have to consider the licensing issues first. Industry analysts predict that indeed smart dust will become pervasive. But will it be Java dust or C/other dust? Consider the above links, dev kit prices, and licenses and then you decide where things are heading. If Sun wants developers to have the choice of pervasiveJava smart dust, something needs to change, and it needs to change soon.  
Kudos to the organizers for making an event like this go smoothly. And, most importantly, there's more depth to the content than I expected. Others I surveyed agreed. You can see the presentations online in the M&E community area. Plus the people here are key company technical personnel, not just marketing heads. So it's a chance to ask some probing questions and get useful answers. Well done!  

The Great Seduction? Blog

Posted by bboyes Jun 12, 2007
Here at the 3rd annual Community Leader's weekend in SFO there are a lot of interesting discussions and techno-gossip. For example, Collabnet just purchased SourceForge. What does this mean for online open-source communities? Are meta-communities still a viable model for the near future? What new features do we want to see at Are podcasts gaining traction in this space? What "best practices" are there for communities? And in my case, how do community leads find the time to keep things fresh? How does anyone? So it should be an interesting weekend, and JavaOne hasn't even started yet...  
"It'll be very good that the Java trap won  

The huge projection screen at the front of the hall seemed to be a cut above what it's been in the past, in terms of brightness, color gamut, and resolution. When will I be able to turn the wall in front of my desk into a giant display? Sun Labs showed me a demo some months ago which mosaics several LCD projectors together to do just that. I'm ready... somehow staring at even a 20-something-inch LCD seems so quaint, really...

Siemens has a Java-enabled cell phone with serial and digital outputs, intended for OEMS to design into their own devices. It looks like a typical development board with easy access to system signals for quick protoyping. If the price is right I want one. This sort of product turns a GSM cell phone into a pluggable network-enabling functional block for hardware which can't attach to ethernet.

GE Healthcare showed some spiffy looking graphics screens of x-ray and CT data, all programmed in Java. It's nice to see that Java's graphics APIs are considered mature and rich enough for a major corporation to design a whole line of mission-critical products around them.

JSR-209 - Advanced Mobile Development - promises to make mobile devices look a lot more like standard J2SE apps with a subset of Java 2D and Swing. This includes "real" threads, I/O, and network stack. Let's face it, the resources in cell phones have come a long way since the J2ME and MIDP specs were conceived, so the hardware capability has been in the phones for a year or two. Phone color screens can now be as good as the best PDAs - in fact, in devices like the Treo, they are one and the same.

JSR1 - Real Time Java - is getting a major enhancement with 2.0, which adds things like realtime garbage collection (based on work at Lund University in Sweden). Threads which need to, can pre-empt the GC, while including provision to be sure those threads don't run out of available heap while doing so. Greg Bollella has been the real-time evangelist at Sun for the last several years. It's good to see Sun recognize the importance of real-time, industrial applications. If Java is to be truly pervasive, it can't just settle for the "80% solution" - it also needs to address the more arcane areas of the market. In reality, real-time systems are almost always connected to non- real-time systems, and you'd like to be able to use the same language and APIs across your entire multi-tier application.

The slot car contest (using real-time Java) was won by a group of German high school students.

Perrone Robotics' "Tommy" Darpa Challenge Vehicle was a big draw, the subject of a Tuesday Technical Session, and won a 2006 Duke award. The vehicle uses real time Java, running on low-cost, off-the-shelf hardware, including two JStamp controllers. Paul's basic approach is "simplicity" and it works very well. They spent $60,000 on the whole project, and went from start to finish in about 10 months. This is a fraction of the time and resources spent by other teams. Tommy 2 is gearing up for the 2007 Urban Challenge. Paul is definitely a "the cup is half full" kind of guy. Sun is now officially one of the sponsors of Perrone Robotics, so Java should be a major contender in the Urban Challenge.


Sony Bails on Robots Blog

Posted by bboyes Jan 27, 2006
This EETimes articleimplies that Sony is not capable of selling $2000 entertainment Aibo pets profitably. Sony will cut 10,000 jobs and pour more money into TVs and high-density DVDs. So much for what many view as a robotic success story: the Aibo US$2000 electronic pets. I don't own one, so it's not clear how this dog "keeps diary with pictures and is able to read e-mail or blog content", but it's a bit surprising that a company with Sony's technical and marketing expertise (rootkits aside) could not make a go of entertainment robots. It's always dangerous to draw general conclusions from a specific incident, so I won't. Other successes remain: the useful Roomba floor sweeper and the Wowee remote control toys, each of which have sold over 1 million. Anyone with some specific insights into why Sony is giving up for now?  
This blog on ZDNET provides links to video and a press release from Global Composites andDistributed Robotics, both in NY state. Here's a PDF press release. If you want a smoother ride, here's a video of an elliptical wheel version. The inventor, Jason Winckler can be seen here with his frog Edna. This may be the same Jason Winclker who plays soccer for Tamarac Middle School. It's an interesting idea, but can spinning a weight around really be more efficient than driving round wheels? The inventors suggest using other forces instead of gravity; perhaps in special applications (where such a force normal to the plane of motion is available) might be very effective. All I can think of right now is the "screen door in a submarine" - where the flow of water perpendicular to the screen would provide a huge force which is easily tapped. So there you have it - the basic concept for a novel submarine screen door cleaner robot. If you make a million from this idea, please send me a check. Similar ideas have been around for a while: driving a square-wheel vehicle over a road composed of inverted catenaries,as seen here. Steering such a vehicle is a bit complicated, since it easily interferes with the synchronization of the wheels. And what easier way to experiment with your own version of this than with Legos(tm)!
On behalf of Systronix and, Bruce Boyes (that would be me) presented a proposal to standardize robotic I/O point tagging in a manner similar to the IEEE1451 STIM (Sensor-Transducer Interface Module). Systronix has developed a robotic tagging API, Java packages, and working examples. This package will soon be posted in the robotics community. It will come as no surprise that there are many ideas about what should be specified, and how. At the end of the presentations, it was proposed that we consider four different task force areas: Middleware - with some sharing concerns about size and footprint Component Profiles - including I/O point tagging Service Profiles Data Structures (Don't ask me exactly what these mean since no is really sure.) This feels like good progress to me, though there is nothing concrete quite yet. There is good interest among active participants, which means that there might actually be people to work on the standards committees. Based on this hopeful sign, OMG approved advancing the Robotis Domain Specific Interest Group to a Domain Task Force. The next meeting is in Feb 2006 in Tampa, and there is still time to submit an RFI response, up until 3 weeks before the Tampa meeting. The robotics community is planning to submit an RFI, based on the I/O point tagging idea, and maybe others. If you have some ideas, you are welcome to join this effort.  
This American Family Physician newsletterreports astonishingly low (to me, anyway) adoption of electronic records systems. Electronic billing and scheduling - yes, but medical records - no. Only 17% of physicians offices use electronic records? And only 8% of physicians use a computerized order entry system, which has the benefits of checking for drug interactions, standard doses, and allergies. Is there an opportunity here or not? Java in combination with open source software has been very successful in scheduling and patient management in Brazil, and in fact, there was a fantastic BOF on this topic at JavaOne in SanFrancisco this past summer. One of the presenters, and the lead developer, Fabiane Bizinella Nardon received a Duke award for this project. There's some interesting discussion about the technology in this ServerSide thread. Brazil is planning to add medical records to this system in the future. So, if there is a need, quantifiable benefits, the technology is available, and has been successfully deployed, what's holding back adoption of such applications? There was quite a bit of discussion of this following the BOF. It's an interesting question...  
This InfoWorld article describes (with no photos) a hand-cranked laptop which one vendor has said could be built at a cost of $110. The machine can be folded in different ways to serve as a computer, electronic book or media player. "We designed the device to perform many roles," said Negroponte, who also heads the One Laptop Per Child nonprofit group. "Learning should be seamless." No details yet on the display, but the laptop will run Linux, and use all open-source software. The idea is to put one into the hands of millions of schoolchildren, especially those in developing countries. Governments must buy 1x10e6 units to participate in the program, which has sponsors including Google. Let's hope Sun gets involved (are they already?) and provides a Java runtime -- what a great platform to showcase the benefits of Java. Here's more, including some sketches of the machine, and the MIT website for the project, including these renderings.  
A $1.35-million contract awarded last week to AeroSat Airborne Internet, LLC of Amherst NH, is the first step in creating an airborne internet in which aircraft act as nodes on a network, passing information on weather, landing conditions and turbulence from one plane to another. The one-year contract calls for Aerosat Airborne Internet to conduct Airborne Internet flight demonstrations on FAA test aircraft at the William J. Hughes Technical Center. Data will be exchanged between the air and ground at 45 megabits per second, then passed from aircraft to aircraft, exhibiting a full range of communications that previously have required the use of satellites. The system is expected to transfer data at 45 megabits per second. It's not clear to me who's doing what. There's the Aerosat company, and the Airborne Internet Consortium. Neither of these entities has anything fresh on their web pages. Then there's this reference at NASA: Transformational Cost Reduction for Airborne Internet Proceedings of the Fourth Integrated Communications, Navigation, and Surveillance (ICNS) Conference and Workshop; August 2004; 17 pp.; In English; See also 20040139058; No Copyright; Avail: CASI; A03, Hardcopy; Available from CASI on CD-ROM only as part of the entire parent document This viewgraph presentation reviews the impact that use of an Airborne Internet will have on the aviation industry. Passenger usage of airborne internet is explored as a possible revenue source for the airlines. AeroSat Corporation, a New Hampshire based company, has commenced work on a project to demonstrate a novel low-cost, broadband, non-satellite communications methodology for aircraft. With nodes moving at 500 mph, with local routers on board, this could give new richness to the idea of a "store and forward" design.  

A recent Time (2005 Oct 24, Technology) article describes the benefits of the 787 Dreamliner, scheduled for production in 2006. First the body: using 50% carbon fiber composite, the 787 has a 20% better fuel burn than similar size jets, but boasts greater range. More interesting, the design focusses on passenger comfort. Cabin pressure will be closer to sea level (meaning the hull has to withstand greater pressurization), and humidity will be in the range of 15-20% instead of today's typical 5%. Windows will use electronic tinting controls instead of pulldown blinds. Will Boeing will follow the lead of smaller GA jets and use solid-state LED lights in the cabin?

According to the Boeing 787 web site, the Dream Liner will boastconstant internet connectivity. We can only hope that it will be included in the price of the ticket and usable during takeoff and landing.

And if you are wondering what device to take to enjoy that connectivity, how about something like the Nokia 770 Internet Tablet?

This blog entry by Marc Wagner was (at least for me) a slap to the forehead. We're all tired of the "Windows vs Linux" non-comparison and that's not my point in blogging his blog. Rather it's these three paragraphs, and the lesson they hold for Embedded Java:Virtually anyone with a high school diploma and few hundred dollars can get certified in Windows server technology in a matter of weeks or months
The NewScientist article describes what appears to be a glaring lack of system design. RFID in a passport might sound like a good idea, but consider this: Without any protection, the passports might be "skimmed"