By Bryan Bergeron
An advantage of using an off-the-shelf robotics platform is that you can focus on the navigation, vision, or other higher level challenges. However, when your robot can’t move in an environment, all the AI in the world isn’t going to help. This was highlighted recently by the accident at the Fukushima nuclear power station in Japan. An underwater robot was used to explore the damage inside the spent fuel pool. Similarly, a robot was used to measure the radiation levels within the facility. Paradoxically, with all of the ‘advanced’ robotics development in Japan, the robot used to explore the facility was a US made iRobot PackBot. The floor was reportedly too cluttered for a Japanese robot to operate.
My take-away lesson from this accident is that for real world robots, you can’t assume a level, smooth, or carpeted working surface. For example, I’ve been experimenting with the Freescale walker robot lately. The walker platform is great for enabling me to work on the electronics, sensors, and the like, but mobility is virtually nill unless the robot is on a smooth desktop. Even a few grains of rice on the desktop can render the walking platform useless.
Have you tried your robot outside? I don’t mean on your smooth driveway, but on an unpaved road or even a dry ditch. Chances are, you’ll need to reconfigure the platform to work within the environment. Even if your robot is based on a four wheel drive R/C truck, you may have to change tire tread to suit the environment.
My experience with underwater robots is limited, but I have tried my hand at a few terrestrial platform designs. For example, I worked with aluminum crawlers (http://www.crustcrawler.com) for a while. Unlike many wheeled platforms, it could overcome modest obstacles, albeit rather slowly. Unfortunately, battery life and payload capacity were limited. It was good for about 100 feet on carpet before requiring a recharge.
The US Army hasn’t given up on multi-legged robot platforms. You’ve probably heard of DARPA’s MULE — a four legged robot that runs on gasoline. It’s not clear when this robot will be ready for active duty. My only experience with a tractor-tread robot is with an upgraded Parallax BoeBot. Like the iRobot PackBot, it is capable of climbing over modest rubble and isn’t phased by small pebbles or uneven surfaces. I’ve even used the tractortread BoeBot outside with reasonable success.
My best outdoor, all-terrain robot is based on a Traxxsas E-Maxx platform. It has lots of power, lots of speed, and great range. The plastic body pops off, leaving plenty of room to mount an aluminum or plastic plate with sensors. Most experimenters that use this platform change the gearing in the transmission to provide a more gentle start and stop, with more control for steering. The $400 E-Maxx is a significant investment, but you’d be hard pressed to come up with such a versatile, waterproof platform from scratch.
Of the commercial, all-terrain, outdoor robot platforms, there’s the Parallax gasoline powered QuadRover. The 89 pound, four-wheeled robot has a 2.5 HP engine. At $1,500, it isn’t cheap, but if you need a real outdoor robot that can carry a significant payload, the QuadRover looks promising. Have an idea for a new or improved platform? If so, let me hear about it. SV