mountaincarrand.txt

CMAC神经网络机械臂控制的设计matlab源码

class MountainCarRand 

Class that implements the randomized variant of the Mountain-Car task, based on the formulation from the book "Reinforcement Learning: An Introdunction" by R. Sutton & A. Borto". The implementation is based on the proposed standard C++ interface classes. This class inherits from the class MountainCar. Can be used as an implementation of the abstract base class Environment. 
Synopsys: #include "mountain_car_rand.h"
Link: mountain_car_rand.cpp
        mountain_car.cpp
        sarepr.cpp
        environment.cpp
        random_numbers.cpp

Public methods: 

class MountainCarRand(double ActVar=0, double PosVar=0) 

Constructor. Creates an instance of the Randomized Mountain Car task, where actions (-1) and (+1) throttle have zero-mean Gaussian noise with variance ActVar and position state variable has zero-mean Gaussian noise with variance PosVar. Both arguments have defualt values zero. With zero variances, the task is equivalent to the original deterministic version. 

void transition(const Action& action, State& s_new, double& r_new, bool& terminal) 

Reimplements the corresponding function of the base class MountainCar and is an implementation of the corresponding pure virtual function of the base abstract class Environment.
Functionality: 
Implements a transtion from CurrentState (base class' data member) in responce to the action specified by action argument. Zero-mean Gaussian noise is added to (-1) and (+1) throttle action and to the position state variable with variance that have been specified as arguments to the constructor of the object; 
updates base class' data members CurrentState, CurrentAction and reward; 
returns values to the agent 
s_new - new state; 
r_new - reward after performed transition; 
terminal - an indication of whether the new state s_new is terminal (boolean true or false value). 

~MountainCarRand() 

Destructor. 

All other functions of the base class MountainCar. 
--------------------------------------------------------------------------------