A Residual Gradient Fuzzy Reinforcement Learning Algorithm for Differential Games

Awheda, Mostafa D.; Schwartz, Howard M.

doi:10.1007/s40815-016-0284-8

Cited by 17 publications

(12 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recently, fuzzy-reinforcement learning methods in [17][18][19][20][21][22] have been proposed to address learning problems in differential games. With these methods, the learning process is achieved by tuning the parameters of two main components, both of which are Fuzzy Inference Systems (FISs) with two sets of parameters (i.e.…”

Section: Motivationmentioning

confidence: 99%

“…If one of the learning algorithms proposed in [17][18][19][20]22] is applied to a multipursuer single-inferior evader PE game there is potential for collisions among the pursuers, particularly if they are near one another or approaching the evader. This motivates the development of a new learning algorithm with the ability to avoid collisions.…”

Section: Motivationmentioning

confidence: 99%

“…In the first example, it is assumed that the evader's movement is based on its DCS, and in the second one, it is assumed that the evader's movement is based on the modified DCS [20]. The final example is similar to the second one, except that the evader can turn with maximum velocity [22]. In all of these example, it is Example 5.1 In this example, the Kalman filter is used to track the evader's movement by estimating its position at each time step.…”

Section: Kalman Filter Model Selectionmentioning

confidence: 99%

“…FACLA is applied to two versions of PE games, and compared through simulation with the FACL [17], a Residual Gradient Fuzzy Actor-Critic Learning (RGFACL) that was proposed in [22] and the PSO-based FLC+QLFIS [140] algorithms; the simulation results were published in [144]…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Learning in the Multi-Robot Pursuit Evasion Game

Al-Talabi¹

View full text Add to dashboard Cite

This thesis proposes different learning algorithms to investigate the learning issues of mobile robots playing differential forms of the Pursuit-Evasion (PE) game. The algorithms are used to reduce: (1) the computational requirements without affecting the overall performance of the algorithm, (2) the learning time, (3) the capture time and the possibility of collision among pursuers and (4) to deal multi-robot PE First and foremost, I am extremely grateful to Almighty Allah, and to Ahlul-Bayt, peace be upon them, for their countless blessings and for the great care they have provided at every moment of my life. Certainly, without their support and bounty, I would never successfully finish this chapter of my life by getting the PhD degree.

show abstract

Section: Motivationmentioning

confidence: 99%

Section: Motivationmentioning

confidence: 99%

Section: Kalman Filter Model Selectionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Learning in the Multi-Robot Pursuit Evasion Game

Al-Talabi¹

View full text Add to dashboard Cite

show abstract

“…In this chapter, we propose a new fuzzy reinforcement learning algorithm for differential games, where the states and actions are continuous. The proposed algorithm is published in [101,102]. The proposed algorithm uses FASs whose parameters are updated differently from the updating mechanisms used in the algorithms proposed in [48,99,100,103,126,127].…”

Section: Introductionmentioning

confidence: 99%

On Multi-Agent Reinforcement Learning in Matrix, Stochastic and Differential Games

Awheda¹

View full text Add to dashboard Cite

In this thesis, we investigate reinforcement learning algorithms on matrix, stochastic, and differential games. In matrix and stochastic games, the states and actions are represented in continuous domains. We propose two decentralized multi-agent reinforcement learning algorithms to solve the problem of learning in matrix and stochastic games when the learning agent has only minimum knowledge about the underlying game and the other learning agents. The proposed algorithms are the constant learning rate-based exponential moving average Q-learning (CLR-EMAQL) algorithm, and the exponential moving average Q-learning (EMAQL) algorithm. We mathematically show that the proposed CLR-EMAQL algorithm converges to Nash equilibrium in games with pure Nash equilibrium. We introduce the concept of Win-or-Learn-Slow (WoLS) mechanism for the proposed EMAQL algorithm so that the proposed algorithm learns fast when it is winning, and learns cautiously when it is losing. We also provide a theoretical proof of convergence to Nash equilibrium for the proposed EMAQL algorithm in games with pure Nash equilibrium. In games with mixed Nash equilibrium, our mathematical analysis shows that the proposed EMAQL algorithm converges to an equilibrium. Although our mathematical analysis does not explicitly show that the proposed EMAQL algorithm converges to Nash equilibrium, our simulation results indicate that the proposed EMAQL algorithm does converge to Nash equilibrium. Our simulation iii I would like to express my deepest gratitude to my advisor, Professor Howard M.

show abstract

An Ensemble Fuzzy Approach for Inverse Reinforcement Learning

Pan

Hwang

et al. 2018

Int. J. Fuzzy Syst.

View full text Add to dashboard Cite

A Residual Gradient Fuzzy Reinforcement Learning Algorithm for Differential Games

Cited by 17 publications

References 46 publications

Learning in the Multi-Robot Pursuit Evasion Game

Learning in the Multi-Robot Pursuit Evasion Game

On Multi-Agent Reinforcement Learning in Matrix, Stochastic and Differential Games

An Ensemble Fuzzy Approach for Inverse Reinforcement Learning

Contact Info

Product

Resources

About