Difference between revisions of "Multi-UAV system design"

From Self-Organization Wiki
Jump to: navigation, search
(What are the design challenges?)
(Group 2)
Line 40: Line 40:
  
 
== Group 2 ==
 
== Group 2 ==
 +
 +
=== Definitions ===
 +
*Autonomy
 +
** Executing tasks within a certain constrained area
 +
** Graduated scale from complete pilot controll to complete uav controll for decission making
 +
** Autonomy can be defined for various scopes ( from the individual robot to a whole system including the operators)
 +
*** a pilot with the UAV could be a system, a tower communicating with them could be external. Where to draw the line is questionable
 +
** usualy the Definition of autonomy only apply to the UAV system itself
 +
** Event hough there are different definitions of autonomy we set the scope only on an individual plattform
 +
 +
=== what is a multi-uav system ===
 +
* UAV:
 +
** Plattform
 +
** (not piloted by a human)
 +
** Capable of aerial Flight
 +
** moves in 3 Dimensions
 +
* n>1
 +
* Can be but does not need to be homogenoeus
 +
* Either there must be a common understanding of the task or some form of communication otherwise it is not a system
 +
 +
=== what are the building blocks of an autonomous multi-UAV system ===
 +
* Hardware requirements
 +
** Power supply
 +
** Communication
 +
** Propulsion
 +
** Sensors
 +
** Computation
 +
* System requirements
 +
** Logic/ Controll logic / AI
 +
** Flight controll system that stablizes the system
 +
* some sort of feedback about the position of sorrounding UAVs is needed for each uav
 +
* Mission control
 +
** Does the mission controll need to know about the UAV systems? Like what type of sensors it uses?
 +
** Depending on definition of the system. If the UAV is autonomous or if the whole system is autonomous
 +
 +
=== what are the design challenges? ===
 +
* Do we need a sense and avoid system for UAVs? (whatever that means)
 +
* Defining the Terms UAV-System and Autonomy
 +
** Where do we draw the line for a system and what falls under Autonomy
 +
* Decicde where and how a system should learn and abstract
 +
* Central vs. Decentral controll
 +
** Central coordination
 +
*** huge communication overhead
 +
*** huge computational constraints
 +
** Decentral design
 +
*** Some form of coordination and parralelization needs to be there
 +
*** Communication is a problem
 +
*** Competing Robots or collaborating?
 +
 +
=== is autonomy desirable? When? ===
 +
 +
* '''Not desirable, it's essential'''
 +
* No conitinouus controll is possible ( communication fails)
 +
* Unpredicted environment changes or internal changes (prop dead)
 +
* Gravity is a problem. if you stand still you drop :)
 +
 +
* multiple UAVs make the problem harder
 +
** As each UAV needs to interact with the environment autonomously it would also need to interact with all the other UAVs (which then are part of the enivornment)
 +
** It would be very hard to controll all of them from a single point
  
 
== Group 3 ==
 
== Group 3 ==

Revision as of 12:54, 8 July 2013

Group 1

Roland Siegwart, Evsen Yanmaz, Torsten Andre, Christof Hoppe, Michael Rischmuller, István Fehérvári, Wilfried Elmenreich

What is a multi-UAV system?

  • airborne
  • two or more UAVs, possibly also heterogeneous
  • have to share knowledge
  • information exchange during operation
  • indirect or direct communication
  • don't have to share a common objective
  • collaboration or coordination

What are the building blocks of an autonomous multi-UAV system?

  • networking, some form of communication
  • sensing system: cameras, ultra-sonic
  • interaction with the environment
  • sharing representation
  • bottom-up/top-down
    • bottom up: team of individuals
    • top down: designed as a team
  • fail-operational or fail-safe
  • swarm of simple systems/swarm with a leader/swarm of highly intelligent systems

What are the design challenges?

  • technical challenges, see above
  • mission-specific or one system fits all?
  • system integration
  • legal issues, regulations
  • responsibility
  • privacy
  • public perception, acceptance

Is autonomy desirable? When?

  • yes, for a large number of UAVs
  • for a large area, remote location
  • if problem is well-defined and easy
  • no, to avoid legal issues
  • if problem is complex, not trackable

Group 2

Definitions

  • Autonomy
    • Executing tasks within a certain constrained area
    • Graduated scale from complete pilot controll to complete uav controll for decission making
    • Autonomy can be defined for various scopes ( from the individual robot to a whole system including the operators)
      • a pilot with the UAV could be a system, a tower communicating with them could be external. Where to draw the line is questionable
    • usualy the Definition of autonomy only apply to the UAV system itself
    • Event hough there are different definitions of autonomy we set the scope only on an individual plattform

what is a multi-uav system

  • UAV:
    • Plattform
    • (not piloted by a human)
    • Capable of aerial Flight
    • moves in 3 Dimensions
  • n>1
  • Can be but does not need to be homogenoeus
  • Either there must be a common understanding of the task or some form of communication otherwise it is not a system

what are the building blocks of an autonomous multi-UAV system

  • Hardware requirements
    • Power supply
    • Communication
    • Propulsion
    • Sensors
    • Computation
  • System requirements
    • Logic/ Controll logic / AI
    • Flight controll system that stablizes the system
  • some sort of feedback about the position of sorrounding UAVs is needed for each uav
  • Mission control
    • Does the mission controll need to know about the UAV systems? Like what type of sensors it uses?
    • Depending on definition of the system. If the UAV is autonomous or if the whole system is autonomous

what are the design challenges?

  • Do we need a sense and avoid system for UAVs? (whatever that means)
  • Defining the Terms UAV-System and Autonomy
    • Where do we draw the line for a system and what falls under Autonomy
  • Decicde where and how a system should learn and abstract
  • Central vs. Decentral controll
    • Central coordination
      • huge communication overhead
      • huge computational constraints
    • Decentral design
      • Some form of coordination and parralelization needs to be there
      • Communication is a problem
      • Competing Robots or collaborating?

is autonomy desirable? When?

  • Not desirable, it's essential
  • No conitinouus controll is possible ( communication fails)
  • Unpredicted environment changes or internal changes (prop dead)
  • Gravity is a problem. if you stand still you drop :)
  • multiple UAVs make the problem harder
    • As each UAV needs to interact with the environment autonomously it would also need to interact with all the other UAVs (which then are part of the enivornment)
    • It would be very hard to controll all of them from a single point

Group 3

What is a multi-UAV system?

  • at least two UAVs
  • combine information

What are the building blocks of an autonomous multi-UAV system?

  • sensing
  • signal processing
  • actuation
  • computation
  • control
  • power
  • power management
  • communication
  • user interface
  • coordination
  • cooperation
  • appropriate mechanical design
  • IN ADDITION: mechanical design and application design blocks

What are the design challenges?

sensor fusion

  • power management
  • processing power
  • redundancy, safety
  • heterogeneity of platforms
  • interdisciplinary knowledge needed
  • strict regulations
  • failure robustness
  • "self healing"


Is autonomy desirable?

  • ... with multi-UAVs, there is no choice
  • better, more deterministic performance

Group 4

Group 5