Jim's
tutorials

Due to the delay in receiving the component parts for Flower-bot, I have directed my focus on the research aspects outlined in the tutorial proposal. I would like to submit 2-3 pages (before Monday, March 3) describing how I think trends in robotics and environmental science will converge within the next decade. For this paper I have looked into several possible robotic and automated applications, their feasibility in terms of available technology, cost, labor, and effectiveness from an ecological perspective. I’m hoping to buy the book listed as resources #1, 2006-2007. I still stand by Kevin Kelly’s book Out of Control as the best starting point for directing my research but I've accumulated several other sources in the mean time. I’ll have about two pages done by today, but I’d rather turn in 3-5 pages with more clearly sited resources (MLA and all that) and more content.

preliminary Paper (needs finishing, today after class ending at 5:30) The Tools and Potential for applied robotics/automation in Environmental research and Ecological Management Liz Mathews

dustry alone has focused billions of dollars on the production of automated-control vehicles, combining first class vision systems with wireless car-to-car and car-to-station communication to ensure maximized driving efficiency and safety. Some robotics companies concentrate on improving universal applications such as visual pattern recognition, sensor quality, or electronic design, while others develop unique ways to incorporate autonomous or intelligent devices into jobs considered too dangerous for on-site human involvement. The spectrum of robotic uses to date covers more territory than anything most science fiction writers could have ever dreamed, but their reign is just beginning in the world of economics, equity, and ecology, and it is last of these which will likely receive more attention in the coming years due to the “Global Warming” and “Peak Oil” scares of the last decade. How can something so obviously unnatural become an unobtrusive part of ecological management? What potential uses might be found for robotics in agriculture, or pollution detection and sequestration, or global atmospheric composition-reading and analysis? These all seem to be fields in need of non-human contributions, and with a little push in the right direction, it’s easy to envision the world’s ecological management systems cooperating with intelligent machines to find smart solutions for many man-made problems. Here’s a brief look at some of the technologies available for commercial use today along with examples of emerging technologies, current projects and applications, and the different viewpoints humanity takes on the growing implementation of robotics in the industrial and ecological environments.

Sensors and Monitoring Monitoring is a crucial and fundamental aspect of all scientific research practices. Unfortunately it is also one of the easiest to botch up in the laboratory due to observational inconsistencies, multi-causal relationships, or poor experiment design. When looking at agriculture for example, a controlled experiment may involve monitoring the progress of one plant species in a given environment, that same plant in a different environment, or the plant’s growth/production when introduced with another plant or animal species. Depending on the experiment’s successful ability to eliminate outside factors, there is always some small level of uncertainty associated with the factors influencing the experimental data. As an experiment grows in complexity, the statistical significance increases (the reliability of the experiment decreases) [1]. Now, most humans reach a pretty low limit when it comes to analyzing very complex systems, but computers can possess a high degree of analytic integrity once customized to a given environment. This makes computational analysis a strong candidate for understanding complex systems like agriculture, regional ecosystems, and global ecologic cycles involving water, atmosphere, minerals and nutrients, soil constitution, species diversity per region etc.

With the right tools these different elements of the natural environment may be measured with accuracy, monitored in a variety of conditions consistently, and provide better precision at the analytic stage of experimentation. Today’s roboticist uses tool kit with a range of sensors to take different measurements within different parameters. Most sensors fall within a family categorization defined by their function or the type of input they read to make a measurement. Here’s a brief list of several sensors used commonly and their general applications.

IR sensors – Infrared sensors use a small pulse of IR electromagnetic waves to detect things like distance from an object, object textures, colors, atmospheric particulate interference, and object movement. IR wavelengths are usually expressed in microns, with the lR spectrum extending from 0.7 to 1000 microns. IR sensors can be used for robot navigation through obstacle courses, or even provide gas and temperature readings by measuring the amount of energy emitted by different surfaces (the 0.7-14 micron band is used for IR temperature measurement) [2]. In more intelligent systems they can provide the data for pattern recognition, or visual memory.

Sound sensors – Respond to distinct changes in vibration frequencies within (or including) the range of human hearing.

Microwave Motion Sensors – These guys detect human presence through motion (or sometimes temperature) readings. The SafePath D38 Microwave Motion Detector uses the patented Human Presence Radar (HPR) to monitor doors and hallways for human traffic. This particular brand (SafePath) allows the sensor to detect separate rates of motion simultaneously. The higher quality version of this sensor incorporates the infrared Floor Reflection Method (FRM) to detect not only localized motion but human presence in higher-density surroundings [3].

Touch Sensors – respond to physical contact between the device and outside objects. These may act as “switches” for human contact with an automated machine (like the touch-sensitive power towel dispensers you encounter in hotels or restaurants), or they might initiate a chain command for a small robot to “back up, reverse, and try a different direction…until you find a direction without immediate obstacles”. Touch sensors can trigger feedback mechanisms and generally have a wide range of potential uses.

Light Sensors (Photometers, photodiodes) – Detect the presence of change in light coming into contact with the sensor. These are a fairly standard addition to many simple robotics projects, and work well when combined with photovoltaic silicon cells to provide increased energy-generation efficiency and power consumption by the robotic device. Plants respond in the same way to subtle changes in shifting sunlight. By adjusting their leaf exposure they maximize the photosynthetic conversion of sunlight to energy needed as fuel for the plant (or a solar power powered robot).

Sensors for Surface Wetness Deposition [4] – There are several ways to measure something like surface wetness (as on plant leaves) using electronic sensors. One way is to place individual electrodes along a grid to measure changes in electrode resistance or capacitance. Simply attaching electrodes directly to a leaf however is not suggested because of the maintenance required in replacing and repositioning the electrodes. (Sutton et al. 1984). Another method is to attach the electrodes to a piece of fabric and apply that to a leaf’s surface, but this method still underestimates the amount of surface water by some degree, and does not do as well when the leaves become dew-saturated.

Speed Sensors – used largely in automotive or transportation vehicles. Acceleration sensors – measure rate of change in velocities (or speed as with the speedometers). “Speed sensors” or “Acceleration sensors” are misleading terms. Both typically rely on one or a combination of other sensor types to produce a reading. Pressure sensors, touch and/or rotation sensors, normal-IR, IR-temperature sensors, or even Micro Motive sensors may be used to determine the relative speed of an object.[5]

Pressure Sensors – Typically used for determining the pressure of a gas or liquid; that is the pressure required to stop a gas or liquid from expanding usually measured as force per unit [6]. Pressure sensors can also be used to determine gas or liquid flow, altitude, and water levels. Some pressure sensors are used more regularly in a constantly changing environment (as in car combustion engines) to keep a steady on any quick changes in pressures. Pressure sensors are particularly useful for aircraft, satellites, weather balloons, and meteorology where both pressure and altitude are relative factors.

Gas Sensors – Gas sensors interact with a gas to initiate the measurement of its concentration. The gas sensor then provides output to a gas instrument to display the measurements such as gas presence and concentration. There’s a whole mess of gas sensors available on the market, but here’s a few thing to consider when picking one for a project: response time (time required to process the signal after contact with gas is made), the maximum distance from a gas source that a sensor can detect the gas, and the flow rate (the minimum speed a gas must be moving across the sensor to elicit a sensor reading)[7]. Gas sensors can detect a large number of gaseous compounds including: ammonia, aerosols, arsine, bromine, carbon dioxide, carbon monoxide, chlorine, dust, fluorine, halocarbons or refrigerants, hydrocarbons, hydrogen, H-chloride and H-cyanide, -selenide and -sulfide, mercury vapor, nitrogen dioxide and other oxides, nitric oxide, organic solvents, ozone, phosphine, sulfur dioxide, and water vapor [8].

Liquid-Level Sensors – measure the difference between liquids and water, or liquids and solids. A constant or point-level reading will alert the integrated system when a distinct level has dropped or risen to a certain point. Just like when your fire alarm goes off letting you know there’s smoke or fire in the house, a liquid-point-level may shut down some part of a larger system, or send a designated alert/alarm. The Liquid-level sensors use data collected from one or a combination of the following sensors: pressure, ultrasound, and vibration.

As we move further down the line of automatic, digital, and mechanical sensors it should start to become obvious that the subtly more complex sensor are really an active combination of sensors similar to the active combination of multiple gates within a circuit or logic structure. The complexity of rules stems from a singular ‘yes’ or ‘no’ response that with time can be designed to branch out in a fashion that answers the right sort of questions as needed. For example:

Question: “Is there a gas leak in the house? Or could my house have a carbon monoxide problem because I haven’t changed the air filters since 1999?” A): this question could be answered using either a very complex array of IR sensors, or a slightly more sophisticated grouping of temperature sensors, or even more organized set of pressure and moisture sensors, but the easiest route to take would be securing one commercial gas sensor designed to detect carbon monoxide or household hydrocarbon gas, or both!

Sources:

[1] Monitoring Plant and Animal Populations

[2] How sensors work “Infrared Temperature Sensors”

[3.] Microwave Motion Sensors and Human Presence Radar http://www.disabilitysystems.com/switches/door_sensors.html

[4.] Cornell University’s comprehensive examination of “Types of Sensors” specifically used for measuring surface wetness deposition (SWD) on plant leafs. http://www.nysaes.cornell.edu/pp/faculty/seem/magarey/leafwet/types.html#eplc

http://en.wikipedia.org/wiki/Pressure_sensor

[7.] Wikipedia on “Gas Sensors” http://sensors-transducers.globalspec.com/LearnMore/Sensors_Transducers_Detectors/Gas_Sensing/Gas_Sensors

[8.] Global Spec on Gas Sensors

Bibliography in alpha order

http://cs.marlboro.edu/ courses/ spring2008/tutorials/ liz/ diary2
last modified Monday March 3 2008 9:46 pm EST