Reinforcement Point and Fuzzy Input Design of Fuzzy Q-Learning for Mobile Robot Navigation System

Pambudi, Arga Dwi; Agustinah, Trihastuti; Effendi, Rusdhianto

doi:10.1109/icaiit.2019.8834601

Cited by 16 publications

(12 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…From the simulation results and experimental results, compared with some other algorithms: Q-learning, DDPG algorithm is better than DQN, Q-learning in terms of value accuracy and control strategy are presented in [13,16], this is also consistent with the DDPG algorithm that the authors have proposed in the paper. Accelerated learning technology and rapid action processing in large environments, can be used to achieve action status maps and meet the mobility needs of mobile robots.…”

Section: Fig 6 Results Of Navigating Robot Turtlebot Done On Rvizsupporting

confidence: 73%

“…Therefore, to build a complete DDPG algorithm, it is always necessary to meet the needs of selecting a control action to a robot, executing the action, receiving rewards, storing and sampling to train the algorithm, calculating of the target function. Subsequently updating the model parameters by minimizing the loss function on all selected samples, followed by selecting the method to update the target neural network parameters, and finally updating the environmental discovery coefficient during the control process [3,5,6,15,16].…”

Section: B the Robot Navigation Using Ddpg Algorithmmentioning

confidence: 99%

See 1 more Smart Citation

Mobile robots interacting with obstacles control based on artificial intelligence

Chuyen¹,

Hoa²,

Dien³

2022

Proceedings of the Sixth International Conference on Research in Intelligent and Computing

View full text Add to dashboard Cite

In this paper, research on the applications of artificial intelligence in implementing Deep Deterministic Policy Gradient (DDPG) on Gazebo model and the reality of mobile robot has been studied and applied. The goal of the experimental studies is to navigate the mobile robot to learn the best possible action to move in real-world environments when facing fixed and mobile obstacles. When the robot moves in an environment with obstacles, the robot will automatically control to avoid these obstacles. Then, the more time that can be maintained within a specific limit, the more rewards are accumulated and therefore better results will be achieved. The authors performed various tests with many transform parameters and proved that the DDPG algorithm is more efficient than algorithms like Q-learning, Machine learning, deep Q-network, etc. Then execute SLAM to recognize the robot positions, and virtual maps are precisely built and displayed in Rviz. The research results will be the basis for the design and construction of control algorithms for mobile robots and industrial robots applied in programming techniques and industrial factory automation control.

show abstract

Section: Fig 6 Results Of Navigating Robot Turtlebot Done On Rvizsupporting

confidence: 73%

Section: B the Robot Navigation Using Ddpg Algorithmmentioning

confidence: 99%

Mobile robots interacting with obstacles control based on artificial intelligence

Chuyen¹,

Hoa²,

Dien³

2022

Proceedings of the Sixth International Conference on Research in Intelligent and Computing

View full text Add to dashboard Cite

show abstract

“…Dalam perencanaan jalur robot dengan hambatan statis, perlu perhitungan yang pasti untuk menghindari hambatan sesuai dengan jalur yang telah direncanakan [4]. Untuk mobil robot yang memiliki heading, penelitian ini fokus pada seberapa optimal robot menghindari hambatan dinamis supaya jalur yang ditempuh adalah minimal [5]. Artikel ini disusun berdasarkan urutan sebagai berikut.…”

Section: Pendahuluanunclassified

Prototipe Robot Avoider sebagai Mesin Penggerak Robot Medical Assistant

Fandidarma

Praditya

Kurniawan

2020

ELECTRA

View full text Add to dashboard Cite

<em>Permasalahan yang diangkat pada penelitian ini adalah membangun prototipe mesin penggerak Robot Medical Assistant yang memiliki fungsi Obstacle Avoidance dengan gerak otonom. Penelitian yang sudah dilaksanakan sebelumnya masih menggunakan joystik untuk menggerakkan robot atau dengan kata lain robot masih secara manual disetir manusia. Karena sangat tidak reliable dan terlalu bergantung pada manusia, maka metode setir manual ini perlu ditingkatkan performanya dengan pengaplikasian metode otonom pada robot. Maka dalam penelitian ini disuguhkan suatu metode yang lebih baik supaya robot dapat bergerak secara otonom dan memiliki fungsi Obstacle Avoidance supaya robot dapat menghindari halangan dengan baik. Jadi user menentukan titik tujuan dimana saja dan tugasnya apa lalu menekan tombol ‘START’ ketika robot siap berangkat. Pengujian dilakukan dengan memberikan perintah untuk robot bergerak dari satu lokasi ke lokasi lainnya dengan dipasang beberapa halangan pada jalur yang tersedia. Dari pengujian ini mendapatkan hasil bahwa robot lumayan berhasil menghindari hambatan dan pergi ke tujuan dengan rerata kesalahan pengukuran sensor sebesar 1.25%.</em>

show abstract

“…There are some Fuzzy adaptations on [24] work like [25] where the Q-functions and action selection strategy are inferred from Fuzzy rules. Also, in order to reduce the number of states needed to shape an MDP model for mobile robots that avoid obstacles, [26] suggests a Fuzzy technique. Because the mobile robot may encounter an infinite number of different conditions.…”

Section: Introductionmentioning

confidence: 99%

Robust Attitude Control of an Agile Aircraft Using Improved Q-Learning

et al. 2022

View full text Add to dashboard Cite

Attitude control of a novel regional truss-braced wing (TBW) aircraft with low stability characteristics is addressed in this paper using Reinforcement Learning (RL). In recent years, RL has been increasingly employed in challenging applications, particularly, autonomous flight control. However, a significant predicament confronting discrete RL algorithms is the dimension limitation of the state-action table and difficulties in defining the elements of the RL environment. To address these issues, in this paper, a detailed mathematical model of the mentioned aircraft is first developed to shape an RL environment. Subsequently, Q-learning, the most prevalent discrete RL algorithm, will be implemented in both the Markov Decision Process (MDP) and Partially Observable Markov Decision Process (POMDP) frameworks to control the longitudinal mode of the proposed aircraft. In order to eliminate residual fluctuations that are a consequence of discrete action selection, and simultaneously track variable pitch angles, a Fuzzy Action Assignment (FAA) method is proposed to generate continuous control commands using the trained optimal Q-table. Accordingly, it will be proved that by defining a comprehensive reward function based on dynamic behavior considerations, along with observing all crucial states (equivalent to satisfying the Markov Property), the air vehicle would be capable of tracking the desired attitude in the presence of different uncertain dynamics including measurement noises, atmospheric disturbances, actuator faults, and model uncertainties where the performance of the introduced control system surpasses a well-tuned Proportional–Integral–Derivative (PID) controller.

show abstract

Reinforcement Point and Fuzzy Input Design of Fuzzy Q-Learning for Mobile Robot Navigation System

Cited by 16 publications

References 4 publications

Mobile robots interacting with obstacles control based on artificial intelligence

Mobile robots interacting with obstacles control based on artificial intelligence

Prototipe Robot Avoider sebagai Mesin Penggerak Robot Medical Assistant

Robust Attitude Control of an Agile Aircraft Using Improved Q-Learning

Contact Info

Product

Resources

About