sarsaagentrt.txt

CMAC神经网络机械臂控制的设计matlab源码

class SarsaAgentRT 
Class that implements an RL learning agent using SARSA(lambda) learning algorithm with replacing eligibility traces. The implementation is based on the proposed standard C++ interface classes. This class implements abstract class Agent. 
Synopsys: #include "sarsart_agent.h"
Link: agent.cpp
        sarsart_agent.cpp
        environment.cpp
        random_numbers.cpp
        sarepr.cpp
        safa.cpp

Public methods: 


SarsaAgentRT(double g, const ActionSet& a_s, StateActionFA* const f, Environment * const e) 

General constructor. Parameters: 
g : Discount facor. 
a_s : Set of available actions. 
f : Pointer to approximation architectures that represent action value functions (one architecture for each action). 
e : Pointer to the Environment object with which the agent will interact. 


setLearningParameters(int argc, char *argv[]) 

Implements the corresponding virtual function of the base Agent class. Sets parameters of the SARSA(lambda) learning algorithm. Parameters are supplied in a command-line argument manner, where argc is the number of parameters in argv array. Each string in argv array is one parameter and its value. Acceptable parameters and their format (to be passed in argv): 
lambda=value : Value of the lambda parameter from [0,1]; 
epsilon=value : Value of the epsilon parameter (in [0,1]) for epsilon-greedy exploration strategy. 
These parameters may be specified as command-line arguments to main(int argc, char *argv[]) and then passed directly to this function. 

static void helpLearningParameters() 

Print out the format for command-line specification of the learning parameters.
Usage: SarsaAgentRT::helpLearningParameters(); 

Non-virtual functions of the base Agent class : 


initTrial 
getRetrun 
getBellmanError 
setArchitectureParameters 
saveArchitectureParameters 


~SarsaAgentRT() 
Destructor. 
--------------------------------------------------------------------------------