CSCE 221 Culture Assignment 4

From Notes
Jump to navigation Jump to search

« previous | Wednesday, Mary 4, 2011 | next »


Localization of multiple unknown transient radio sources using multiple paired mobile robots with limited sensing ranges

Chang-Young Kim, Dezhen Song, Yiliang Xu

Hostile sensor network search and destroy

Localization:

  • triangulation
  • time of arrival
  • Radio frequency localization (received signal strength)

Problems we have with Hostile networks (vs. freindly) Signal source anonymity Short transmission duration Dynamic transmission patterns Unknown network size Perspective limit of robots

Constraints:

  • LImited sensing range
  • Variable transmission powers

Assumptions

  • Centralized control
  • Free 2D space
  • Short and light network traffic
  • Circular radiation pattern
  • Stationary radio sources

Deriving probability of location

Sensor layer (should output probability), SPOG, and Motion planner


Three types of paired robots:

  • Both receive signal
  • one receives signal
  • Neither receive signal (no information)


Product of information from different sources at different time causes probability to converge.

Pair distance must be far enough to triangulate source, but close enough to overlap a little

  • PRWA triangulated fastest



Approximate Characterization of Multi-Robot Swam "shapes" in Sublinear Time

Lantau Liu, Klappi, et al.


Can an individual know the shape of the swarm?

  • How long/wide is the network (geodesic diameter)?
  • How many hops (geodesic distance) are necessary to get from one end to the other.

Problems:

  • Distributed (everyone depends on everyone else)
  • Local sensing and communication
  • For a very large swarm size , becomes too expensive

Vocabulary: Sublinear: less than Path distance: sum of edge length along path


Boundary detection:

  • outer edge of swarm (peripheral agents) Measuring angles between individual and surrounding robots (obtuse)
  • inside swarm (inner agents)
  1. initiate root agent on periphery
  2. recognize and count peripheral agents
  3. Follow trace of 1st traversal
  4. Traverse inner agents
  5. ...

Send a message to the middle of the swarm, pass it along the most "tangent" angle between robots to other side; count number of "hops" made

This is sublinear since the peripheral agents initiate the message, and the inner agents only transmit the message

Simulation on dense (high degree) and sparse (low degree) uniform distribution of agents




Utilizing Roadmaps in Evacuation Planning

Sam Rodriguez

Complex reaction and many agents; Trying to get to an exit

Small to large map-based algorithms for simple office buildings

Similar to pursuit-evasion problem: communication, sharing, environment knowledge

Agents don't have to exactly follow the path. Path can be optimized and simplified.

Agents can be directors (avoid this exit, avoid this area, etc)