George Anwar of Integrated Motion uses Vision Tools for Electronic Nose Application
One cool application is the Exoskeleton. I read about this a year ago but didn’t think NI tools would play much of a role. I was wrong. George applied data acquisition to the application. For those who are not familiar with the Exoskeleton it is a device that fits around a person’s body and allows the wearer to carry heavy loads with less effort. The “exoskeleton” shell holds most of the weight. For some high resolution photos from the Berkeley effort, please check out this web-site.
There are two versions of the exoskeleton, passive and fully powered. George worked on the fully powered one. He designed a data acquisition and network system to collect data from the sensors on the robot and move it to the computer strapped to the back of the user. If you wired all the sensors, you would end up with a wire harness of 120 cables on the wearer’s waist. So they implemented a high speed network that reduced the cabling to six wires. They used FireWire cable with a PC-104 computer. It was able to transmit at 100 MHz. They were going to design cRIO into it but the developers moved to a passive unit—which didn’t use power. It’s driven by prosthetics based on a passive knee. It changes the metabolic rate of how one walks. Students can carry 150 lb backpacks effortlessly. You get tired, but it’s a different type of fatigue. It uses small muscle motion rather than large muscle motion.
George is also working on a data mining project that’s still in the discovery stage. He is working with University California researcher, Arun Majumdar who is trying to create polymer-specific receptors that sense explosive materials. This work creates a pool of data that must be searched and analyzed. They used a MEMS-based device which has 64 cantilevers on it. The target material makes a deposit on a receptor of the cantilever when it reacts. The stress of the binding causes a trigger which a camera monitors. The combination of multiple signals indicates the chemical composition. You can read more about how they apply this research here. George Anwar uses NI Vision’s Cameralink tools to test it. The cantilever array looks like a series of camera pixels. He built a vision package that runs cases and stores images of the chemical reactions.
“You’re not looking for a specific peak compound, but rather you are looking for a series of peaks that may be more unique,” George said. The original work comes under the heading of Electronic Nose technology. I blogged about it last year. Electronic noses use a sensor array with pattern matching software.
“Analytically we don’t know what to look for,” he said. Vision is the closest NI product technology to solving this area because it works with an array of pixels which can represent an array of sensors. Also, the temperature must be controlled as it can alter the results.
Pattern matching is the key to making it work. George indicates that he can see some of the tools that can help such as NIs vision software. They are still seeking the key compounds. If you want to detect a particular array, and you have an array of 64 elements, you may want to position the array element in a diagonal pattern so if the chemical exists, you can look for that line. Or you can group the arrays such that if a chemical is detected, it turns on the receptors creating a circle of a particular size. These are some of the techniques, but it’s not optimal. Right now, they are getting some “hits” but it’s a painful task of working through the process of correlating the hits to the chemical.
George thought NIs vision tools could be applied to MEMs applications since a camera can be thought of as an array of sensors. This opens up the door for applying Vision tools to a whole new range of applications. These sensors are analog which is ideal for a vision camera to measure and the software to analyze. You can see more about George’s work at his web site.