Cooperative Control of UAVs

1
Cooperative Control of UAVs

• A mixed-initiative approach

Ltn. Elói Pereira Portuguese Air Force
Academy E-mail etpereira_at_emfa.pt
2
Summary

• AFA project on UAVs
• Cooperative Control of UAVs in mixed initiative
  environments
• Military and Civil applications
• Formalism for Allocation and Exchange vehicles
  within teams
• Example Load Balancing between teams
• Testbed description
• Conclusions and future work.

3
ANTEX Portuguese Air Force Project on UAVs

• Development of UAVs platforms to use as
  technologies demonstrators in several fields as
• Scientific Research
• Defense
• Civil applications
• Give Air Force know how in operation of UAVs
• To promote RD initiatives with others
  organizations
• Faculty of Engineering of Porto University
• Technical Lisbon University
• University of California at Berkeley
• University of Victoria
• University FAF Munich
• Ecoles d'officiers de l'Armée de l'air
  (internship of two cadets)

4
Cooperative Control of UAVs

• Vehicles exchange information and commands in a
  network, changing their dependencies, states and
  mission roles to achieve a common goal

Source MICA Project
5
Mixed Initiative

• Planning procedure and execution control must
  allow intervention by experienced human
  operators.
• Essential experience and military insight of
  these operators cannot be reflected in
  mathematical models
• It is impossible to design vehicle and team
  controllers that can respond satisfactorily to
  every possible contingency. In unforeseen
  situations, these controllers ask the human
  operators for direction. Pravin et al.

The Commander is an actuator
Plant
Better Performance
Better Decisions
Better Info
Commander/ Operator
Battlespace
Decision Control
Decision Aids
Courses of Action
Embedded Hierarchy
Better status knowledge
Measured status
Estimation
Source MICA Project
6
Military and Civil Applications

• Research topic that has been attracting the
  attention of control, communications and computer
  science researchers
• Possible applications with large societal impact
  are raising interest outside the scientific
  community
• Military missions
• Combat
• Reconnaissance
• Surveillance
• Patrol
• Civil missions
• Forest inspections
• Security
• Environmental applications

7
Example Strike Enemy Air Defenses (SEAD) mission

• MICA Mixed-Initiative Control of Automata Teams
  (DARPA)
• Mission Attack of the Blue force of UAV against
  Red's ground force of SAM sites and radars

Primary targets
sms14
sls7
sls8
sms11

• Maneuvers
• Follow_path
• Loitter
• Attack_jam

sms17 sls6
sms15
sms12
sls5
sms13
Blue base
J. Borges de Sousa, T. Simsek e P. Varaiya, Task
planning and execution for UAV teams,
Proceedings of the Decision and Control
Conference, Bahamas, 2004
8
Example

• Execution control

Primary targets
Team A
sms14
Leg 6
sls7
sls8
sms11
Leg 1
sms17 sls6
sms15
Team B
sms12
sls5
sms13
Blue base
J. Borges de Sousa, T. Simsek e P. Varaiya, Task
planning and execution for UAV teams,
Proceedings of the Decision and Control
Conference, Bahamas, 2004
9
Example

• Execution control

Primary targets
Attack segment
sms14
Leg 6
sls7
sls8
sms11
Leg 1
sms17 sls6
sms15
sms12
sls5
sms13
Attack segment
Blue base
J. Borges de Sousa, T. Simsek e P. Varaiya, Task
planning and execution for UAV teams,
Proceedings of the Decision and Control
Conference, Bahamas, 2004
10
Example

• Execution control

Primary targets
Attack segment
sms14
Leg 6
Leg 7
sls7
sls8
Precedes
Safe path
sms11
Leg 1
sms17 sls6
Leg 2
sms15
sms12
Safe path
sls5
sms13
Blue base
Attack segment
J. Borges de Sousa, T. Simsek e P. Varaiya, Task
planning and execution for UAV teams,
Proceedings of the Decision and Control
Conference, Bahamas, 2004
11
Example

• Execution control

Primary targets
Subtask 2
Leg 8
sms14
Leg 6
Leg 7
sls7
sls8
Precedes
sms11
Leg 1
Subtask 1
Leg 5
sms17 sls6
Leg 4
Leg 2
sms15
Leg 3
sms12
sls5
sms13
Blue base
12
Formalism for Allocation and Exchange vehicles
within teams
teams
vehicles

• Matrix formalism
• Initial Allocation of vehicles to teams
• Transition-vehicle incident matrix
• Final Allocation of vehicles to teams
• The formalism could be used to design high level
  controllers in mixed-initiative environments

Decision variables
Team-transition incident matrix
13
Load-balancing algorithm

• Load-balance the number of vehicles within teams
• Heterogeneous vehicles
• Different fuel reserves
• Different number of weapons
• Different types of payloads
• Performance Measure Difference between the
  number of vehicles in the team and the number of
  vehicles initially planned for that team
• Problem is solved as a Binary Integer Programming
  (BIP) optimization problem.

14
Example Load-balancing

• Five teams with different necessities
• Fuel constraints

15
Actual UAV system configuration
Autopilot Avionics
Sensors
Ground Station
Payload Devices
Servos
Neptus Command and Control Interface (FEUP)
Autopilot Avionics
Sensors
Payload Devices
Servos
16
Advanced Configuration - Work in progress

• Autopilot manages low-level flight control
• PC-104 for higher-level tasks (vision processing,
  trajectory planning, coordinated between UAVs)

Sensors
Servos
Autopilot Avionics
Payload Devices
PC-104
Ground Station
Sensors
Servos
Autopilot Avionics
Payload Devices
PC-104
Aircraft Low level control and logging
Payload High level Control and logging
17
Vehicles
ANTEX-X02 (AFA)
Silver Fox (ACR)
NOVA (AFA)
Flying Wing (AFA)
ANTEX-X03 (AFA)
Lusitânia (FEUP)
18
Operation of UAVs and Cooperative control
simulation
19
Conclusions and future work

• Cooperative control of UAVs is a research field
  with large margin of progression and with
  possible applications with societal impact (dull,
  dirty and dangerous missions)
• The intervention of the operator in the planning
  and execution control (mixed-initiative) is
  crucial in missions with large uncertainty,
  namely in military operations
• ANTEX developments in a near short term
• Operation with several UAVs
• Track and follow structures (rivers, roads)
  based on vision payloads
• Autonomous landing
• Mid-term objective
• Operation with others types of unmanned vehicles
  (underwater, surface).

20
Questions?
Thank you for your attention
21
Vehicles Characteristics

• Lusitânia UAV (FEUP)
• Maximum Take Off Weight 10 kg
• Wing Span 2.4 m
• Payload 5 kg
• Endurance 0.75h
• On board payload wireless video camera
• Nova UAV (AFA)
• Maximum Take Off Weight 4 kg
• Wing Span 1.6 m
• Payload 0.5 kg
• Endurance 0.75h
• Flying Wing UAV (AFA)
• Maximum Take Off Weight 3 kg
• Wing Span 1.6 m
• Payload 0.2 kg
• Endurance 0.3h

• ANTEX-M X03 (AFA)
• Wing Span 6 m
• Maximum Speed 130 km/h
• Stall Speed 40 km/h
• Maximum Take Off Weight 100 kg
• Payload 30 kg
• Engine 22 hp
• Endurance/Fuel Capacity 0.3h/4L
• ANTEX-M X02 (AFA)
• Maximum Take Off Weight 10 kg
• Wing Span 2.4 m
• Maximum Speed 151 km/h
• Payload 4 kg
• Endurance/Fuel Capacity 0.3h/0.2L
• Silver Fox (ACR)
• Maximum Take Off Weight 12.2 kg
• Wing Span 2.4 m
• Maximum Speed 203 km/h
• Payload 2.27 kg