Grant: Discovery Grants Program - Individual
Carleton University
$20,000.00
Adaptive and intelligent control of multiple cooperative robots
FY 2013-14
Summary
The objective of this research is to investigate methods of adaptive and learning control for multiple mobile robots. Such systems will be very useful in search and rescue operations wherein a number of unmanned autonomous vehicles search for lost individuals or can identify environmental hazards. These systems will be useful in securing border areas and industrial complexes. The proposed research focuses on how the robots will learn to work together. I am proposing innovative methods of multi robot and multi agent reinforcement learning. To focus the research further I will investigate two scenarios. In one case the robots will play the multi pursuer evader game and in the other case they will play the guarding a territory game. I am proposing original methods of multi robot learning in both cases. In the first stage of research I will focus on the discrete grid world. In the grid world I will propose new methods of learning and then I am going to extend these ideas for use on real robots. In the continuous domain I am proposing a new structure for multi robot learning based on an actor-critic learning structure. Furthermore, the actor-critic structure is based on a fuzzy inference system to estimate the state action value function over the continuous domain. This is also a new formulation. A particularly innovative idea that I am proposing is the use of heterogeneous vehicles that behave differently as though they had their own personalities. Much of the research will be done using Matlab as a simulation platform. The researchers will be using our own experimental robot facilities. We have developed an experimental facility that uses up to six mobile robots working together. We have tested the swarming and tracking algorithm and the evader/pursuer game. The researchers will use this experimental facility for experimentally evaluating the performance. I will also be collaborating with the Royal Military College, Defence Research and Development Canada and Allen Vanguard and the researchers will have access to their experimental facilities as well.
Program
Discovery Grants Program - Individual
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
NSERC's Awards Data
As published
Application id
37216-2011
Name nom
Schwartz, Howard
Department département
Systems and Computer Engineering
Organization id
23
Institution établissement
Carleton University
Province en
Ontario
Province fr
Ontario
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2013
Competition year année de concours
2011
Award amount
20000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
1512
Committee name en
Mechanical Engineering
Committee name fr
Génie mécanique
Area of application code
513
Area of application group en
Manufacturing processes and products
Area of application group fr
Procédés et produits industriels
Area of application en
Electrical and electronic machinery and equipment (including computer hardware)
Area of application fr
Machinerie et équipement électriques et électroniques (y compris matériel informatique)
Research subject code
2109
Research subject group en
Mechanical engineering
Research subject group fr
Génie mécanique
Research subject en
Mechanical systems and instrumentation
Research subject fr
Instrumentation et systèmes mécaniques
Partie
3
Nb partie
5
Application title
Adaptive and intelligent control of multiple cooperative robots
Keyword
NSERC
Application summary
The objective of this research is to investigate methods of adaptive and learning control for multiple mobile robots. Such systems will be very useful in search and rescue operations wherein a number of unmanned autonomous vehicles search for lost individuals or can identify environmental hazards. These systems will be useful in securing border areas and industrial complexes. The proposed research focuses on how the robots will learn to work together. I am proposing innovative methods of multi robot and multi agent reinforcement learning. To focus the research further I will investigate two scenarios. In one case the robots will play the multi pursuer evader game and in the other case they will play the guarding a territory game. I am proposing original methods of multi robot learning in both cases. In the first stage of research I will focus on the discrete grid world. In the grid world I will propose new methods of learning and then I am going to extend these ideas for use on real robots. In the continuous domain I am proposing a new structure for multi robot learning based on an actor-critic learning structure. Furthermore, the actor-critic structure is based on a fuzzy inference system to estimate the state action value function over the continuous domain. This is also a new formulation. A particularly innovative idea that I am proposing is the use of heterogeneous vehicles that behave differently as though they had their own personalities. Much of the research will be done using Matlab as a simulation platform. The researchers will be using our own experimental robot facilities. We have developed an experimental facility that uses up to six mobile robots working together. We have tested the swarming and tracking algorithm and the evader/pursuer game. The researchers will use this experimental facility for experimentally evaluating the performance. I will also be collaborating with the Royal Military College, Defence Research and Development Canada and Allen Vanguard and the researchers will have access to their experimental facilities as well.