[Lab Bench Online]
Doing The Robot, And More, At ESC 2008
Robots, USB, micros, multimedia, boards, software, hands-on demos...what's not to like about ESC 2008?
William Wong
|
ED Online ID #20107 |
November 11, 2008
Finally heading back home by train towards Philly—and the Phillies just wrapped up the World Series last evening—so that should spice up my pass through the city of brotherly love a little. It is definitely nicer traveling home via train than plane. A five minute wait at the train station, and a five hour ride with lots of leg room and an AC outlet plus great scenery is definitely the way to go.
I’m going to run through most of what I saw in reverse-order so I can get to the fun stuff first—mainly the keynote by James McLurkin that I stuck around for—more on that in just a minute.
First I’d like to make a quick note about attendance. Things seemed on par with last year, although traffic on the show floor was slow at times due to a few sessions close to each other that had heavy attendances. Some changes on food locations and other critical details are likely to change when ESC rolls around again. The feedback from the vendors was positive overall, so some of you kicking the tires are still buyers.
Robots I was surprised by the attendance at Thursday’s keynote. This is the day after the show floor shut down, so robotics definitely draws crowds. Dr. James McLurkin (Fig. 1) was the speaker. I’ve known James for many years and was expecting a neat presentation. He exceeded my expectations and had a small cadre surrounding him after things wrapped up, and everyone had questions rolling off their tongues, including myself.
The presentation was about robots and covered quite a bit of ground. But the bulk of it centered on his work with swarm robotics. This was not your usual PowerPoint presentation, although there were plenty of slides and videos. Onlookers must have been amused by the swarm of robots rolling around the stage floor doing their master’s bidding (Fig. 2). It was entertaining and informative. There were the usual snafus, but the small, almost cube-like robots did their part. They are fun to play with (Fig. 3).
The demos included a simple start with follow-the-leader to more interesting bubble sorts using only localized behaviors. Another demo pulled eight programmers and engineers from the audience to highlight convergence techniques using calculators and interaction between the participants. Now if they could just have done it as quickly as the middle school kids did.
One aspect of his talk was to encourage the audience to support local science fairs, schools, and robotic events to foster science and engineering in our youth. This is something I wholeheartedly endorse. If you happen to be in the Mercer County, New Jersey area shoot me an e-mail. I run the Mercer Science and Engineering Fair and could use your help.
But back to robots. One question I did ask was about debugging swarms. As expected, the chore is not easy and there is a lot of work to do in this area. The current crop of robots is programming in C, but Java is the target for future work. “Why”, you might ask? Academia at its worst? No. Program correctness. James stressed that iterative testing and making sure the code was as correct as possible makes a big difference.
I could ramble on about more robotic tricks, photos, and stories, but we need to hit more highlights on the other technologies and products that were at ESC.
USB One thing that we included on our videos for Electronic Design and Engineering TV were a few companies that are making more use of USB inside the box.
Small Form Factor SIG (SFF-SIG) announced their MiniBlade (Fig. 4) that is a 40-pin socket with interfaces like USB, PCI Express (PCIe), and SATA. It will likely be a USB storage socket, but it has the potential to handle much more since these are generic interfaces. SiliconSystems has its StorageBlade flash memory that fits into the locking Samtec socket.
Another force at the show was the StackableUSB group, started by Micro/sys. They use a different Samtec plug and socket that has half a dozen interfaces including USB, serial peripheral interface (SPI), and I2/sup>C. Micro/Sys was showing off a PC/104 size motherboard and quarter-sized peripheral boards. The USB peripheral boards also have mini-USB socket so you can use the boards with any USB host via a standard cable.
WinSystems has some peripheral boards as well for the Pico I/O standard (Fig. 5) from the SFF-SIG. The peripheral board plugs into the new Pico ITX board from VIA Technologies. I’ll be looking at this new standard more closely in the future, but it answers the biggest issue with the original Pico ITX board, expansion. The SUMIT connectors on the board provide a stackable collection of PCIe, USB, and a host of other interfaces including LPC.
AccesIO foregoes any new standard and sticks with the normal USB connector on a wide range of I/O boards from real relays to very fast analog boards. The boards are PC/104-size, so they can be easily mounted in a stack; in boxes designed for PC/104 boards; or anywhere you can put mounting holes. The latter is key since the peripheral board can be placed away from the host and near the device. This can impact everything from size to cooling, allowing a designer much more flexibility than a stack, rack, or other existing solutions. USB hub vendors are going to love these—including the new $350 Fast USB Digital Waveform Generator module.
Toradex is now delivering very small USB sensors like this three-axis device (Fig. 6). It has a range of these Oak USB Sensors that work nicely with any USB host, as well as the new Atom-based Woodpecker board they announced at the show or the Marvell PXA-based Colibri DIMM-style modules. These support Windows CE and Linux.
USB has significant limitations but it often exceeds the requirements for a wide range of applications. One thing to look out for in the far future (my guess is a year or two) will be this same type of remote architecture but using PCIe instead of USB. Wired PCIe is now used for cabling boxes, but lower cost cables and connectors inside a box will allow distribution of I/O but with faster connections. It could even be more effective than CAN or Ethernet within a box. Just a thought.
One discussion I will bring up in the future is one that I had with many of the vendors about generating standards for USB peripherals. The base USB standards are the starting point for all of these devices but once above things like a USB serial connection programmers must turn to custom protocols. A standard mechanism for interacting with devices like general-purpose I/Os (GPIOs), digital-to-analog converters (DACs), and analog-to-digital converters (ADCs) like ZigBee does with its profiles would make this type of support really take off.
Another idea I have been tossing around is the inclusion of a simple virtual machine/interpreter to run scripting language programs on these USB devices. Everyone of the devices has a microcontroller on them and rarely is 100% of the flash or memory used up. Adding a simple interpreter comes at little or no cost. It would allow generic applications to be downloaded to standard devices like a DAC, where it could perform functions such as averaging or logging. Any takers on an open source project?
Reprints
