Q-learning is a one of the most popular and frequently used model-free reinforcement learning method. Among the advantages of this method is independent in its prior knowledge and there is a proof for its convergence to the optimal policy. One of the main limitations of More
Q-learning is a one of the most popular and frequently used model-free reinforcement learning method. Among the advantages of this method is independent in its prior knowledge and there is a proof for its convergence to the optimal policy. One of the main limitations of this method is its low convergence speed, especially when the dimension is high. Accelerating convergence of this method is a challenge. Q-learning can be accelerated the convergence by the notion of opposite action. Since two Q-values are updated simultaneously at each learning step. In this paper, adaptive policy and the notion of opposite action are used to speed up the learning process by integrated approach. The methods are simulated for the grid world problem. The results demonstrate a great advance in the learning in terms of success rate, the percent of optimal states, the number of steps to goal, and average reward.
Manuscript profile
Path planning of mobile robot is one of the most important topics in mobile robotic discussion. The aim of this study is to find a continuous path from an initial position to the final target; So that, it should be free of collision and optimal or near to optimal. Since More
Path planning of mobile robot is one of the most important topics in mobile robotic discussion. The aim of this study is to find a continuous path from an initial position to the final target; So that, it should be free of collision and optimal or near to optimal. Since path planning problem of robot is one type of optimization problems, the evolutionary algorithms can be used to solve this problem. Nowadays, clonal selection algorithm is frequently used to solve the problems because of having valuable computational characteristics. But very little attempts have been done in the field of using this method to solve robot path planning problem. Few accomplished attempts are actually a kind of improved genetic algorithm. In this research, an efficient method for robot path planning in the presence of obstacles is designed using all the features of the clonal selection algorithm. The proposed method is evaluated in various environments with different runs in terms of the proposed path length criteria and the number of generations needed to generate the path. Based on the results of experiments, the proposed method shows better performance than the genetic algorithm in all environments and all the evaluation parameters. Especially, by increasing the number of obstacles vertices and also concave obstacles, the proposed method shows much more efficient performance than the genetic algorithm. Also, comparing the performance of the proposed method with the BPSO algorithm (presented in another study) indicates the superiority of path planning algorithm based on the clonal selection.
Manuscript profile