"Arise, go forth and conquer as of old." - Sir Bedivere's classic entreaty to King Arthur immediately came to mind when we got our first look at the new SRV-1 mobile robotic webcam from the Surveyor Corporation. What we saw quickly opened our eyes to all sorts of exciting possibilities and opportunities to apply the technology, so before long we were digging into the robot's details and background with its creator, Howard Gordon.
Video cameras are great. Being able to expand our vision, and our horizons beyond the confines of our office or home, and to experience the rest of the world is fantastic.
It's even more fantastic and exciting when you start to think about cameras and vision systems that are mobile - robots equipped with video cameras that can "explore, discover, and report" to borrow a very apt phrase from the Surveyor website.
And, if you extend the metaphor just a little bit further, then you begin to realize that it won't belong before we see interactive, inter-communicating robot swarms.
Howard, the founder and CEO of Surveyor, really believes in that scenario. He has the requisite experience and background with robotics, software, vision systems, and webcams. And, he's set off on a journey to make that vision a reality.
Here's how our dialog evolved so far-
Lem: Could you explain the SRV-1 concept, and how you happened to come up with it?
Howard: The concept is to build a low-cost vision-based wireless computing platform with wheels, and I'm not entirely certain where it will lead, but I find the concept personally exciting ... it's something I've given a fair amount of thought to over the past 30+ years. I actually developed one of the first pan/tilt object tracking systems back around 1990 (engineers from Sony and Canon checked it out), but didn't get any backing, so I got sidetracked into developing JPEG/MPEG software, putting radio and TV stations on the Internet, etc, and only recently came back to my robotics interests.
Lem: I think the early development prototypes included an on-board compass, but that seems to have disappeared somewhere along the way. What happened?
Howard: The on-board compass proved to be almost completely useless as an indoor navigation aid. There was no easy way to provide enough separation from the robot motors on such a small platform, not to mention other indoor magnetic fields. So we replaced the compass with an array of infrared detectors that can be used to locate beacons as well as for inter-robot communication.
The controller module includes IR detectors for location and robot to robot communication.
Lem: We've seen some other robot platforms that are focused primarily on home security applications and have a more "closed - don't hack me" feel to them. What type of customers and applications do you think will the first to adopt the SRV-1 robot platform and technology?
Howard: I think in the early stages, the target customers are techno-geek hobbyists, educators, and researchers (Lego Mindstorms and Parallax Boe-Bot owners probably fit this profile), though there is hope that some commercial volume applications as well as some OEM licensing opportunities will make themselves obvious and accessible.
Lem: The SRV-1 robot base is certainly solid and well designed. Have you considered using other designs or bases?
Howard: We sell each of the modules separately and would encourage people to use them with their own bases if they like. We have a couple of projects in the works for testing the SRV-1 controller with other mobile bases including a radio controlled battle tank.
Radio controlled Abrams battle tank
The tank has lots of available space to mount
the SRV-1 controller and camera module
It's not difficult to imagine some multi-player game scenarios with these. I expect we will be able to interface to the turret controls. I'm actually working on code now for inter-robot communication using the new infrared components and thinking about how to develop some basic swarm behaviors.
We also have a blimp project in the works. The blimp has only 3 motors, so interface should be simple. I thought about trying this with the Silverlit X-UFO, but expect that will be a lot more difficult to control.
I'll post lots of photos and information in our online Surveyor Robotics Journal for the blimp, tank and other projects as they progress so people can follow along and adopt the ideas for their own SRV-1 projects.
Lem: Those are great! I can see how they could be a lot of fun. Do you plan to setup the tank to shoot pellets?
Howard: I actually think it would be more interesting to use the IR sensors and emitter for a "laser tag" type of game, perhaps also adding some sound effects. Longer term, I'm looking at how to develop swarm behavior using the inter-robot communication mechanism, which would be interesting for games and other applications. Lots of possibilities ...
Lem: What about the speed and responsiveness from a video and remote control perspective?
Howard: With regard to any of this, I'm thinking "slow motion". A large part of the idea is to develop cooperating packs of autonomous robots requiring minimal user intervention, so the laser tag is just a means of keeping score if the format is some kind of game.
Given the relatively low data rates (and corresponding low video frame rates / resolution), we have to be realistic about the scope of applications we can target. Real-time remote flight control, racing, etc, would be a lot of fun, but non-kludge system would require a pretty good radio supporting 1.5+ megabits/second and an MPEG1 / MPEG2 compression engine, all of which starts to cost real money.
Lem: We've seen some low end, toy-market products announced recently with video capability, like an R/C car with a video feed. How do they compare with the SRV-1?
Howard: The R/C car you mentioned operates in the analog domain, so the resulting video isn't pulled into a computer or accessible via internet, but it will be at full video frame rates with good resolution, it certainly looks like a fun product. The first SRV robots I prototyped used that type of wireless camera for testing, with a kludge to capture the analog video and convert it to digital.
It is a challenge to base a robotics system on compressed digital video such as we're doing with SRV-1. You probably haven't seen many l ow-cost robotic systems with digital video support, as these systems are really limited by bandwidth and computational requirements.
But once video data has moved to the digital domain, there are big advantages in transport, storage and manipulation. From the start, my "vision" for the SRV-1 robot was to have video as a sensor for interpreting surroundings, and for multiple SRV robots to share this sensory data. Because the robots have digital radio links back to base stations that are Internet connected, it's natural to provide webcam functionality and teleoperation, but long-term, this is only a side benefit of having autonomous wireless mobile bots with vision.
The human hand makes an excellent and readily
available target for the robot to track...
Lem: So what's your most recent exciting SRV-1 development?
Howard: Well, we just did some testing with object recognition and tracking. The robot is rotating, and the motion is a little jerky, but the tracking actually works pretty well. I need to add forward and reverse motion to follow the object as it gets closer or more distant, plus more sophisticated object identification, but it's not a bad start. The really cool thing is that it took no more than 15 lines of additional code to implement the tracking. Here's a video clip showing our initial testing:
Howard has been gracious enough to provide us with an evaluation SRV-1, so we'll be doing some extensive testing and experimenting of our own. Of course we'll post about the results of our 'hacking' the SRV-1 here on Robots Dreams.
Robotics Journal - details on the SRV-1 projects, progress, and lots of detail including photos
Surveyor Corporate Website - Includes other interesting webcam products and customer profiles
Idylls of the King - The Passing of Arthur Alfred Lord Tennyson (1869)