Distributed Intelligence in Autonomous Robotics
Exam 2 Study Guide
CS594, Section 30682, Spring 2003
Last Update: March 10 (no more updates planned)
- Exam 2 will be in class on Thursday, March 13th. It will be closed
book, closed notes.
- All material covered in class Tuesday, February 11th
through Thursday, March 6th, will be on
- Readings 10 and 12-17 will be covered on the exam. BE SURE TO READ ALL
Use the following questions as a guide for studying for Exam #2.
There is no guarantee that these questions will be the ones asked on the
exam, but they are a good indicator of the type of questions to expect.
- What is the difference between explicit and implicit communication?
- Give 2 examples of implicit communication in nature.
- In defining a communications taxonomy, what are 3 "dimensions"
of communication that are important to characterize?
- What were the fundamental findings of Balch and Arkin in comparing
no communication, state communication, and goal communication?
- Given a particular formation and formation control strategy, define
the sensing and communications requirements.
- Give an example of a formation that is determined by constraints
other than distance and orientation.
- What formation type would you recommend for (a) minimizing penetration
through a barrier, (b) maximizing surface area coverage per unit time,
(c) moving as a group through an obstacle field?
- Define a metric for evaluating robot team performance in formation-keeping
(specify mathematically, using precise terms).
- Assume you have a genetic algorithm that can learn behaviors of
predators and prey for a herding application (like the Werner and Dyer paper
we read). Describe the effect/impact of learning the behaviors of
predators and prey
simulataneously. Discuss whether or not the two types of behaviors could be
learned separately. If so, how; if not, why not.
- Show how Mataric's basis behaviors (homing, safe-wandering, dispersion,
aggregation, following) and higher-level group behaviors (flocking and
surrounding) can be combined to generate herding.
- Assume you want to build an approach using potential fields that
enables multiple robots
to track multiple moving targets (similar to paper #14 by Parker). What would
be the source(s) of attractive potential field(s) in this application?
What would be the source(s) of the repulsive potential field(s) in this
- If you have many more targets than robots, how would you enable
robots to put higher preference on targets that are not being observed by
any other robot?
- Describe a baseline program against which you would evaluate your
potential-field based target tracking program.
- How would the potential-field based
target tracking approach need to be adapted to use it for tracking multiple
targets on the main floor of Claxton Complex?
- Reconfigurable Robots:
- What is the difference between lattice-type reconfigurable robots
and chain-type reconfigurable robots?
- In the CONRO approach to reconfigurable robots, how does a module know
what effect a hormone should have on it?
- In a lattice-type system, show which modules would "scrunch" and which
would expand in order to move a module through the crystal from a specified
starting position to a specified goal position.
- Describe the high-level process of the grow-melt algorithm.
- What is the purpose of the "grain" concept in Rus's Crystalline Robots?
- Given a set of modules and a set of operations, show what the
final state of the system will be (for either the CONRO hormone approach or
the lattice-type approach, which will be specified).
- Path Planning / Traffic Management:
- For what types of applications is a pre-planning process for multi-robot path
generation important, and for what types of applications is a traffic management
approach more appropriate?
- In the multi-robot path planning approach of Guo and Parker (paper #17),
what are the two major planning steps?
- Describe what a coordination diagram represents and how it is used in the
multi-robot path planning process. What are the axes of a coordination diagram
and what do they represent?
- What does a path from the origin to the point at the "far corner" (i.e.,
the point where all values along all axes are maximized) represent?
- List 4 traffic rules that would be useful for multiple robots operating
along a fully automated interstate highway system (i.e., only robot vehicles
- Provide a set of rules that would prevent deadlock
if 4 robots arrive at a 4-way stop sign at the same time.
- Write pseudocode to generate a specific distributed robot behavior
(the specific behavior will be specified).