rklee Uber-Karma
Joined: 11 Sep 2003 Posts: 495 Location: Malvern, PA Pittsburgh, PA
|
Posted: Thu Oct 28, 2004 9:38 pm Post subject: Summary of Special Seminar by Karl Hedrick UC Berkeley |
|
|
Here's a summary I wrote for a distinguished guest speaker seminar I had at my school (requirement for all graduate students). Feel free to comment it's a really fun and interesting topic I think.... This was somehting I had strongly considered for my own research (using ALICE 2000 robots).
Graduate Seminar I
Report #2
Reginald Lee rklee@andrew.cmu.edu
Seminar: Multiple Unmanned Vehicle Control
Speaker: Karl Hedrick, UC Berkeley
Date: 10-22-04
Carnegie Mellon University
Summary:
Amid national security concerns and the war on terrorism, unmanned land, air, and sea vehicles have been researched heavily in this country. It is common knowledge that the Predator unmanned aerial surveillance vehicle has been used in the war on terrorism. These unmanned vehicles can be a great benefit to our security forces at home and abroad. The pilot of these vehicles is either a computer system or just a controller sitting in a safe location. This reduces the amount of danger and risk to our troops.
Karl Hedrick?s research focuses primarily on the use of multiple unmanned vehicles. Using multiple vehicles has many benefits depending on the application. Several uses that were listed by Karl Hedrick were: autonomy, collaboration, distribution, and decentralization. This allows for the benefits of low maintenance and high returns, performance of more complex functions through multitasking, can survive losses of individual vehicles without failing the mission, and could allow for lower cost vehicles. Karl Hedrick?s research has been applied to ground traffic, aerial helicopters for surveillance in dangerous locations by collaborating with a ground convoy, for search and rescue coverage over large bodies of water, and with reconfigurable floating runways.
One issue that the research focused on was the method for controlling of these multiple vehicles. Hedrick?s research showed that when vehicles only track their relative position to the vehicles surrounding it, that errors tend to propagate and amplify as you progress down the chain. He explained it to be similar to the long wait that drivers experience in traffic at a stop light. All the cars must wait until the car in front of it moves before accelerating which causes a long wait for cars in the rear of the line despite the light being green. Hedrick solved this problem and created ?mesh stability? by having each vehicle consider extra information such as the relative position to a lead vehicle. Hedrick showed that by making use of this extra information, a convoy of vehicles was able to form much closer formations and maintain the initial relative positions much better. Hedrick also experimented with using even more information such as the location of the lead vehicle, the orientation, and the formation shape. This was especially useful when considering a group of aerial vehicles flying in a formation that must change shape due to the environment. A simulation by Hedrick showed the success of such a control method.
An interesting issue that was discussed by Karl Hedrick was the difference between a top down and a bottom up approach to accomplishing a task. His work primarily focuses on the top down approach, where he had a hybrid system of controls that from a very complex and controlled method, accomplishes the task. He explained that one of his students has pursued the bottom up approach. This approach is analogous to how ants work together to accomplish a task. A very simple set of instructions controls a swarm of these vehicles to accomplish a task. Similarly, ants have a very basic set of instructions but are able to accomplish fairly complex tasks. This bottom up approach was simulated by his student using a fairly simple set of instructions that included a ?to do? list, a ?done? list, and a ?doing? list as well as parameters for communications between nearby vehicles. This simulation proves the feasibility and potency of this method. _________________ "My Heart Is In the Work" - Andrew Carnegie |
|