CSCI 8970 – Colloquium Series – Fall 2010 – Second Event
Robotic Sensors Networks for Environmental Monitoring
Monday, September 20, 2010
|
Presenter: |
Volkan Isler |
During this interesting session, Volkan Isler introduced to us what his research team at the UMN has recently been researching and implementing. Through the use of a sensor network, Isler has programmed a robotic boat to help with a carp control program by monitoring carps within a particular region through the use of wireless controllers. The boat was first developed and launched in 2008. While the original boat stayed within his via points, the boat was somewhat fragile and the boat’s router kept breaking. As a result, in 2010, the team introduced its second drone. This time, the robot has not only a sparkling and sharp look, but it was also resistant to the wake generated by nearby boats. Carps have increasingly threatened to enter the great lake system and various control alternatives are being researched. To prevent a recent invasion, the Chicago sanitary system poisoned the ship canal to prevent carps from entering the system. This project hopes to offer an alternative to unselective destructive control measures. The new boat covered 1.23 kilometers in the single trial coverage. Regarding some of the technical details, the boat’s antenna pans in steps of 15 degrees every 5 seconds, and the fishes beep every second. Despite their success, they are currently working to make the robot completely autonomous.
Another project in which Isler is involved includes using roller robots as data mules. Prof. Cavender-Bares of the University of Minnesota used a Cambell CR3000 Micrologger to collect environmental data in hopes of finding out how topical plant populations respond to climate change. The data used to be collected manually. Now, however, through the use robots as data collectors, they can cover a large area without using relay nodes. The robots have been tried at the park between 15th Ave and Como, as well as in the flats behind Coffman Union.
During the presentation, Isler engaged the audience by providing various algorithm problems such as: Given locations of sensors, what is a good order to visit them? The class went through the algorithm logic together taking into account the relationship between the distance to the node, the travelling time and how long it takes to download the data. Among the results, we saw how the minimum path is of length at least 0.48r. They are currently trying to improve the energy use by the system.
Isler also introduced us to include various polygon pursuit evasion games. He argued that no matter what three cops are enough to capture a criminal within any polygon, and if it is a polygon with no obstacles, a single pursuer is enough. He is currently exploring of what happens if the cops cannot see the robbers at all times. Another project he is working on includes a robotic videoconferencing system to record realistic setting such as when a person is playing with their pets or walking around the house. Currently, multiple robots are needed to have an image at all times. All in all, we can all agree that despite the challenges, robotic sensor networks provide an almost limitless potential for their implementation when fused with a bit of creativity.
