rlSACAgent

Soft actor-critic (SAC) reinforcement learning agent

Since R2020b

Description

The soft actor-critic (SAC) algorithm is an actor-critic off-policy method for environments with discrete, continuous, and hybrid action-spaces. The SAC algorithm attempts to learn the stochastic policy that maximizes a combination of the value of the optimal policy and its entropy. The policy entropy is a measure of policy uncertainty given the state. A higher entropy value promotes more exploration. Maximizing both the reward and the entropy balances exploration and exploitation of the environment. A soft actor-critic agent uses two critics to estimate the value of the optimal policy, while also featuring target critics and an experience buffer. SAC agents support offline training (training from saved data, without an environment), as well as training with evolutionary strategies.

Note

Soft actor-critic agents with an hybrid action space do not support training with an evolutionary strategy. Also, they cannot be used to build model based agents. Finally, while you can train offline (from existing data) any SAC agent, only SAC agents with continuous action space support batch data regularizer options.

For more information, see Soft Actor-Critic (SAC) Agent. For more information on the different types of reinforcement learning agents, see Reinforcement Learning Agents.

Creation

Syntax

agent = rlSACAgent(observationInfo,actionInfo)

agent = rlSACAgent(observationInfo,actionInfo,initOptions)

agent = rlSACAgent(actor,critics)

agent = rlSACAgent(___,agentOptions)

Description

Create Agent from Observation and Action Specifications

agent = rlSACAgent(observationInfo,actionInfo) creates a SAC agent for an environment with the given observation and action specifications, using default initialization options. The actor and critics in the agent use default deep neural networks built using the observation specification observationInfo and action specification actionInfo. The ObservationInfo and ActionInfo properties of agent are set to the observationInfo and actionInfo input arguments, respectively.

agent = rlSACAgent(observationInfo,actionInfo,initOptions) creates a SAC agent with deep neural networks configured using the specified initialization options (initOptions).

Create Agent from Actor and Critic

agent = rlSACAgent(actor,critics) creates a SAC agent with the specified actor and critic objects and default agent options.

Specify Agent Options

agent = rlSACAgent(___,agentOptions) sets the AgentOptions property for any of the previous syntaxes.

Input Arguments

initOptions — Agent initialization options

rlAgentInitializationOptions object

Agent initialization options, specified as an rlAgentInitializationOptions object.

actor — Actor

rlContinuousGaussianActor object | rlDiscreteCategoricalActor object | rlHybridStochasticActor object

Actor that implements the policy, specified as one of the following:

• An rlContinuousGaussianActor object (for continuous action spaces)

• An rlDiscreteCategoricalActor object (for discrete action spaces)

• An rlHybridStochasticActor object (for hybrid action spaces)

For more information on creating actor approximators, see Create Policies and Value Functions.

Note

A continuous SAC agent automatically reads the action range from the UpperLimit and LowerLimit properties of the action specification (which is used to create the actor), and then internally scales the distribution and bounds the action. Therefore, when creating the network for an rlContinuousGaussianActor for use in a SAC agent, do not add a tanhLayer as the last nonlinear layer in the mean output path. If you bound the mean value output directly (for example by adding a tanhLayer right before the output), the agent does not calculate the entropy of the probability density distribution correctly. Note that you must still add a softplus or ReLU layer to the standard deviations path to enforce nonnegativity. For more information, see Soft Actor-Critic (SAC) Agent.

critics — Critic

rlQValueFunction object | two-element row vector of rlQValueFunction objects | rlVectorQValueFunction object | two-element row vector of rlVectorQValueFunction objects

Critic, specified as one of the following:

• For continuous action space SAC agents — a vector containing one or two rlQValueFunction objects.

• For discrete or hybrid action space SAC agents — a vector containing one or two rlVectorQValueFunction objects.

Note

When you have two critics, they must be unique objects with the same observation and action specifications. The critics can either have different structures or the same structure but with different initial parameters.

For more information on creating critics, see Create Policies and Value Functions.

Properties

ObservationInfo — Observation specifications

specification object | array of specification objects

Observation specifications, specified as an rlFiniteSetSpec or rlNumericSpec object or an array containing a mix of such objects. Each element in the array defines the properties of an environment observation channel, such as its dimensions, data type, and name.

If you create the agent by specifying an actor or critic, the value of ObservationInfo matches the value specified in the actor and critic objects. If you create a default agent, the agent constructor function sets the ObservationInfo property to the input argument observationInfo.

You can extract observationInfo from an existing environment, function approximator, or agent using getObservationInfo. You can also construct the specifications manually using rlFiniteSetSpec or rlNumericSpec.

Example: [rlNumericSpec([2 1]) rlFiniteSetSpec([3,5,7])]

ActionInfo — Action specification

rlNumericSpec object | rlFiniteSetSpec object | vector containing one rlFiniteSetSpec followed by one rlNumericSpec object

Action specification, specified as one of the following:

• One rlNumericSpec object (for continuous action spaces)

• One rlFiniteSetSpec object (for discrete action spaces)

• A vector consisting of one rlFiniteSetSpec followed by one rlNumericSpec object (for hybrid action spaces)

The action specification defines the properties of an environment action channel, such as its dimensions, data type, and name.

Note

For non-hybrid action spaces (either discrete or continuous) or only one action channel is allowed. For hybrid action spaces, you must have two action channels, the first one for the discrete part of the action, the second one for the continuous part of the action.

If you create the agent by specifying an actor and critic, the value of ActionInfo matches the value specified in the actor and critic objects. If you create a default agent, the agent constructor function sets the ActionInfo property to the input argument ActionInfo.

You can extract actionInfo from an existing environment, function approximator, or agent using getActionInfo. You can also construct the specification manually using rlFiniteSetSpec and rlNumericSpec.

Example: rlNumericSpec([2 1])

AgentOptions — Agent options

rlSACAgentOptions object

Agent options, specified as an rlSACAgentOptions object.

If you create a SAC agent with default actor and critic that use recurrent neural networks, the default value of AgentOptions.SequenceLength is 32.

ExperienceBuffer — Experience buffer

rlReplayMemory object | rlPrioritizedReplayMemory object | rlHindsightReplayMemory object | rlHindsightPrioritizedReplayMemory object

Experience buffer, specified as one of the following replay memory objects.

• rlReplayMemory (default)

• rlPrioritizedReplayMemory

• rlHindsightReplayMemory

• rlHindsightPrioritizedReplayMemory

Note

Agents with recursive neural networks only support rlReplayMemory and rlHindsightReplayMemory buffers.

During training the agent stores each of its experiences (S,A,R,S',D) in the buffer. Here:

• S is the current observation of the environment.

• A is the action taken by the agent.

• R is the reward for taking action A.

• S' is the next observation after taking action A.

• D is the is-done signal after taking action A.

The agent then samples mini-batches of experiences from the buffer and uses these mini-batches to update its actor and critic function approximators.

UseExplorationPolicy — Option to use exploration policy for simulation and deployment

true (default) | false

Option to use exploration policy when selecting actions during simulation or after deployment, specified as a one of the following logical values.

• true — Use the base agent exploration policy when selecting actions in sim and generatePolicyFunction. Specifically, in this case the agent uses the rlStochasticActorPolicy policy with the UseMaxLikelihoodAction property set to false. Since the agent selects its actions by sampling its probability distribution, the policy is stochastic and the agent explores its action and observation spaces.

• false — Force the agent to use the base agent greedy policy (the action with maximum likelihood) when selecting actions in sim and generatePolicyFunction. Specifically, in this case the agent uses the rlStochasticActorPolicy policy with the UseMaxLikelihoodAction property set to true. Since the agent selects its actions greedily the policy behaves deterministically and the agent does not explore its action and observation spaces.

Note

This option affects only simulation and deployment; it does not affect training. When you train an agent using train, the agent always uses its exploration policy independently of the value of this property.

SampleTime — Sample time of agent

1 (default) | positive scalar | -1

Sample time of the agent, specified as a positive scalar or as -1.

Within a MATLAB^® environment, the agent is executed every time the environment advances, so, SampleTime does not affect the timing of the agent execution.

Within a Simulink^® environment, the RL Agent block that uses the agent object executes every SampleTime seconds of simulation time. If SampleTime is -1 the block inherits the sample time from its input signals. Set SampleTime to -1 when the block is a child of an event-driven subsystem.

Note

Set SampleTime to a positive scalar when the block is not a child of an event-driven subsystem. Doing so ensures that the block executes at appropriate intervals when input signal sample times change due to model variations.

Regardless of the type of environment, the time interval between consecutive elements in the output experience returned by sim or train is always SampleTime.

If SampleTime is -1, for Simulink environments, the time interval between consecutive elements in the returned output experience reflects the timing of the events that trigger the RL Agent block execution, while for MATLAB environments, this time interval is considered equal to 1.

This property is shared between the agent and the agent options object within the agent. Therefore, if you change it in the agent options object, it gets changed in the agent, and vice versa.

Example: SampleTime=-1

Object Functions

train	Train reinforcement learning agents within a specified environment
sim	Simulate trained reinforcement learning agents within specified environment
getAction	Obtain action from agent, actor, or policy object given environment observations
getActor	Extract actor from reinforcement learning agent
setActor	Set actor of reinforcement learning agent
getCritic	Extract critic from reinforcement learning agent
setCritic	Set critic of reinforcement learning agent
generatePolicyFunction	Generate MATLAB function that evaluates policy of an agent or policy object

Examples

Create Default SAC Agent

Create environment and obtain observation and action specifications. For this example, load the environment used in the example Compare DDPG Agent to LQR Controller. The observation from the environment is a vector containing the position and velocity of a mass. The action is a scalar representing a force, applied to the mass, ranging continuously from -2 to 2 Newton.

env = rlPredefinedEnv("DoubleIntegrator-Continuous");
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

The agent creation function initializes the actor and critic networks randomly. Ensure reproducibility by fixing the seed of the random generator.

rng(0)

Create a SAC agent from the environment observation and action specifications. Since actInfo is an rlNumericSpec object, rlSACAgent creates an agent with a continuous action space. When actInfo is an rlFiniteSetSpec object, rlSACAgent creates an agent with a discrete action space.

agent = rlSACAgent(obsInfo,actInfo);

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension)})

ans = 1x1 cell array
    {[0.0546]}

You can now test and train the agent within the environment. You can also use getActor and getCritic to extract the actor and critic, respectively, and getModel to extract the approximator model (by default a deep neural network) from the actor or critic.

Create Default Discrete SAC Agent

Create an environment with a continuous action space and obtain its observation and action specifications. For this example, load the environment used in the example Train PG Agent with Custom Networks to Control Discrete Double Integrator. The observation from the environment is a vector containing the position and velocity of a mass. The action is a scalar representing a force, applied to the mass, ranging continuously from -2 to 2 Newton.

env = rlPredefinedEnv("DoubleIntegrator-Discrete");
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create an agent initialization option object, specifying that each hidden fully connected layer in the network must have 128 neurons.

initOpts = rlAgentInitializationOptions(NumHiddenUnit=128);

The agent creation function initializes the actor and critic networks randomly. Ensure reproducibility by fixing the seed of the random generator.

rng(0)

Create a SAC agent from the environment observation and action specifications using the initialization options. Since actInfo is an rlFiniteSetSpec object, rlSACAgent creates an agent with a discrete action space. When actInfo is an rlNumericSpec object, rlSACAgent creates an agent with a continuous action space.

agent = rlSACAgent(obsInfo,actInfo,initOpts);

Extract the deep neural network from the actor.

actorNet = getModel(getActor(agent));

Extract the deep neural networks from the two critics. Note that getModel(critics) only returns the first critic network.

critics = getCritic(agent);
criticNet1 = getModel(critics(1));
criticNet2 = getModel(critics(2));

Display the layers of the first critic network, and verify that each hidden fully connected layer has 128 neurons.

criticNet1.Layers

ans = 
  6x1 Layer array with layers:

     1   'input_1'       Feature Input     2 features
     2   'fc_1'          Fully Connected   128 fully connected layer
     3   'relu_body'     ReLU              ReLU
     4   'fc_body'       Fully Connected   128 fully connected layer
     5   'body_output'   ReLU              ReLU
     6   'output'        Fully Connected   3 fully connected layer

Plot the networks of the actor and of the second critic, and display the number of weights.

plot(actorNet)

summary(actorNet)

   Initialized: true

   Number of learnables: 17.2k

   Inputs:
      1   'input_1'   2 features

plot(criticNet2)

summary(criticNet2)

   Initialized: true

   Number of learnables: 17.2k

   Inputs:
      1   'input_1'   2 features

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension)})

ans = 1x1 cell array
    {[2]}

You can now test and train the agent within the environment.

Create Default SAC Agent Using Initialization Options

Create an environment and obtain its observation and action specifications. For this example, load the environment used in the example Compare DDPG Agent to LQR Controller. The observation from the environment is a vector containing the position and velocity of a mass. The action is a scalar representing a force, applied to the mass, ranging continuously from -2 to 2 Newton.

env = rlPredefinedEnv("DoubleIntegrator-Continuous");
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Create an agent initialization option object, specifying that each hidden fully connected layer in the network must have 128 neurons.

initOpts = rlAgentInitializationOptions(NumHiddenUnit=128);

The agent creation function initializes the actor and critic networks randomly. Ensure reproducibility by fixing the seed of the random generator.

rng(0)

Create a SAC agent from the environment observation and action specifications using the initialization options. Since actInfo is an rlNumericSpec object, rlSACAgent creates an agent with a continuous action space. When actInfo is an rlFiniteSetSpec object, rlSACAgent creates an agent with a discrete action space.

agent = rlSACAgent(obsInfo,actInfo,initOpts);

Extract the deep neural network from the actor.

actorNet = getModel(getActor(agent));

Extract the deep neural networks from the two critics. Note that getModel(critics) only returns the first critic network.

critics = getCritic(agent);
criticNet1 = getModel(critics(1));
criticNet2 = getModel(critics(2));

Display the layers of the first critic network, and verify that each hidden fully connected layer has 128 neurons.

criticNet1.Layers

ans = 
  9x1 Layer array with layers:

     1   'concat'        Concatenation     Concatenation of 2 inputs along dimension 1
     2   'relu_body'     ReLU              ReLU
     3   'fc_body'       Fully Connected   128 fully connected layer
     4   'body_output'   ReLU              ReLU
     5   'input_1'       Feature Input     2 features
     6   'fc_1'          Fully Connected   128 fully connected layer
     7   'input_2'       Feature Input     1 features
     8   'fc_2'          Fully Connected   128 fully connected layer
     9   'output'        Fully Connected   1 fully connected layer

Plot the networks of the actor and of the second critic, and display the number of weights.

plot(actorNet)

summary(actorNet)

   Initialized: true

   Number of learnables: 17.1k

   Inputs:
      1   'input_1'   2 features

plot(criticNet2)

summary(criticNet2)

   Initialized: true

   Number of learnables: 33.6k

   Inputs:
      1   'input_1'   2 features
      2   'input_2'   1 features

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension)})

ans = 1x1 cell array
    {[-0.9867]}

You can now test and train the agent within the environment.

Create SAC Agent from Actor and Critics

Create an environment and obtain observation and action specifications. For this example, load the environment used in the example Compare DDPG Agent to LQR Controller. The observation from the environment is a vector containing the position and velocity of a mass. The action is a scalar representing a force, applied to the mass, ranging continuously from -2 to 2 Newton.

env = rlPredefinedEnv("DoubleIntegrator-Continuous");
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

Define bounds on the action. The SAC agent automatically uses these values to internally scale the distribution and bound the action properly.

actInfo.LowerLimit=-2;
actInfo.UpperLimit=2;

SAC agents use two Q-value function critics. A Q-value function critic takes the current observation and an action as inputs and returns a single scalar as output (the estimated discounted cumulative long-term reward for taking the action from the state corresponding to the current observation, and following the policy thereafter).

To model the parameterized Q-value function within the critics, use a neural network with two input layers (one for the observation channel, as specified by obsInfo, and the other for the action channel, as specified by actInfo) and one output layer (which returns the scalar value). Note that prod(obsInfo.Dimension) and prod(actInfo.Dimension) return the number of dimensions of the observation and action spaces, respectively, regardless of whether they are arranged as row vectors, column vectors, or matrices.

Define each network path as an array of layer objects. Assign names to the input and output layers of each path. These names allow you to connect the paths and then later explicitly associate the network input layers with the appropriate environment channel.

% Observation path
obsPath = [
    featureInputLayer(prod(obsInfo.Dimension),Name="obsInLyr")
    fullyConnectedLayer(32)
    reluLayer
    fullyConnectedLayer(16,Name="obsPathOutLyr")
    ];

% Action path
actPath = [
    featureInputLayer(prod(actInfo.Dimension),Name="actInLyr")
    fullyConnectedLayer(32)
    reluLayer
    fullyConnectedLayer(16,Name="actPathOutLyr")
    ];

% Common path
commonPath = [
    concatenationLayer(1,2,Name="concat")
    reluLayer
    fullyConnectedLayer(1)
    ];

% Assemble dlnetwork object.
criticNet = dlnetwork;
criticNet = addLayers(criticNet,obsPath);
criticNet = addLayers(criticNet,actPath);
criticNet = addLayers(criticNet,commonPath);

% Connect layers
criticNet = connectLayers(criticNet,"obsPathOutLyr","concat/in1");
criticNet = connectLayers(criticNet,"actPathOutLyr","concat/in2");

Plot the network.

plot(criticNet)

To initialize the network weights differently for the two critics, create two different dlnetwork objects. You must do this because the agent constructor function does not accept two identical critics.

criticNet1 = initialize(criticNet);
criticNet2 = initialize(criticNet);

Display the number of weights.

summary(criticNet1)

   Initialized: true

   Number of learnables: 1.2k

   Inputs:
      1   'obsInLyr'   2 features
      2   'actInLyr'   1 features

Create the two critics using the two networks with different weights and the names of the input layers. Alternatively, if you use exactly the same network with the same weights, you must explicitly initialize the network each time (to make sure weights are initialized differently) before passing it to rlQValueFunction. To do so, use initialize.

critic1 = rlQValueFunction(criticNet1,obsInfo,actInfo, ...
    ActionInputNames="actInLyr", ...
    ObservationInputNames="obsInLyr");

critic2 = rlQValueFunction(criticNet2,obsInfo,actInfo, ...
    ActionInputNames="actInLyr", ...
    ObservationInputNames="obsInLyr");

For more information about value function approximators, see rlQValueFunction.

Check the critics with a random observation and action input.

getValue(critic1,{rand(obsInfo.Dimension)},{rand(actInfo.Dimension)})

ans = single

-0.1330

getValue(critic2,{rand(obsInfo.Dimension)},{rand(actInfo.Dimension)})

ans = single

-0.1526

SAC agents use a parameterized stochastic policy, which for continuous action spaces is implemented by a continuous Gaussian actor. This actor takes an observation as input and returns as output a random action sampled from a Gaussian probability distribution.

To approximate the mean values and standard deviations of the Gaussian distribution, you must use a neural network with two output layers, each having as many elements as the dimension of the action space. One output layer must return a vector containing the mean values for each action dimension. The other must return a vector containing the standard deviation for each action dimension.

The SAC agent automatically reads the action range from the UpperLimit and LowerLimit properties of actInfo (which is used to create the actor), and then internally scales the distribution and bounds the action.

Therefore, do not add a tanhLayer as the last nonlinear layer in the mean output path. If you bound the mean value output directly (for example by adding a tanhLayer right before the output), the agent does not calculate the entropy of the probability density distribution correctly. Note that you must still add a softplus or ReLU layer to the standard deviations path to enforce non-negativity. For more information, see Soft Actor-Critic (SAC) Agent.

Define each network path as an array of layer objects, and assign names to the input and output layers of each path.

% Define common input path.
commonPath = [
    featureInputLayer(prod(obsInfo.Dimension),Name="obsInLyr")
    fullyConnectedLayer(32)
    reluLayer(Name="comPthOutLyr")
    ];

% Define path for mean value.
meanPath = [
    fullyConnectedLayer(32,Name="meanPthInLyr")
    reluLayer
    fullyConnectedLayer(16)
    reluLayer
    fullyConnectedLayer(prod(actInfo.Dimension),Name="meanOutLyr")
    ];

% Define path for standard deviation.
stdPath = [
    fullyConnectedLayer(32,Name="stdPthInLyr")
    reluLayer
    fullyConnectedLayer(16)
    reluLayer
    fullyConnectedLayer(prod(actInfo.Dimension))
    softplusLayer(Name="stdOutLyr")
    ];

% Assemble dlnetwork object.
actorNet = dlnetwork;
actorNet = addLayers(actorNet,commonPath);
actorNet = addLayers(actorNet,meanPath);
actorNet = addLayers(actorNet,stdPath);

% Connect layers.
actorNet = connectLayers(actorNet,"comPthOutLyr","meanPthInLyr/in");
actorNet = connectLayers(actorNet,"comPthOutLyr","stdPthInLyr/in");

Plot the network.

plot(actorNet)

Initialize network and display the number of weights.

actorNet = initialize(actorNet);
summary(actorNet)

   Initialized: true

   Number of learnables: 3.2k

   Inputs:
      1   'obsInLyr'   2 features

Create the actor using actorNet, the observation and action specification objects, and the names of the input and output layers.

actor = rlContinuousGaussianActor(actorNet, obsInfo, actInfo, ...
    ActionMeanOutputNames="meanOutLyr",...
    ActionStandardDeviationOutputNames="stdOutLyr",...
    ObservationInputNames="obsInLyr");

For more information about continuous Gaussian actors approximators, see rlContinuousGaussianActor.

Check your actor with a random input observation.

getAction(actor,{rand(obsInfo.Dimension)})

ans = 1x1 cell array
    {[-0.8205]}

Specify training options for the critics.

criticOptions = rlOptimizerOptions( ...
    Optimizer="adam", ...
    LearnRate=1e-3,... 
    GradientThreshold=1, ...
    L2RegularizationFactor=2e-4);

Specify training options for the actor.

actorOptions = rlOptimizerOptions( ...
    Optimizer="adam", ...
    LearnRate=1e-3,...
    GradientThreshold=1, ...
    L2RegularizationFactor=1e-5);

Specify agent options, including training options for actor and critics.

agentOptions = rlSACAgentOptions;
agentOptions.SampleTime = env.Ts;
agentOptions.DiscountFactor = 0.99;
agentOptions.TargetSmoothFactor = 1e-3;
agentOptions.ExperienceBufferLength = 1e6;
agentOptions.MiniBatchSize = 32;

agentOptions.CriticOptimizerOptions = criticOptions;
agentOptions.ActorOptimizerOptions = actorOptions;

Create the SAC agent using actor, critics, and options.

agent = rlSACAgent(actor,[critic1 critic2],agentOptions)

agent = 
  rlSACAgent with properties:

        ExperienceBuffer: [1x1 rl.replay.rlReplayMemory]
            AgentOptions: [1x1 rl.option.rlSACAgentOptions]
    UseExplorationPolicy: 1
         ObservationInfo: [1x1 rl.util.rlNumericSpec]
              ActionInfo: [1x1 rl.util.rlNumericSpec]
              SampleTime: 0.1000

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo(1).Dimension)})

ans = 1x1 cell array
    {[0.4490]}

You can now test and train the agent within the environment.

Create SAC Agent Using Custom Recurrent Neural Networks

For this example, load the environment used in the example Compare DDPG Agent to LQR Controller. The observation from the environment is a vector containing the position and velocity of a mass. The action is a scalar representing a force, applied to the mass, ranging continuously from -2 to 2 Newton.

env = rlPredefinedEnv("DoubleIntegrator-Continuous");
obsInfo = getObservationInfo(env);
actInfo = getActionInfo(env);

SAC agents use two Q-value function critics. To model the parametrized Q-value function within the critics, use a recurrent neural network, which must have two input layers one output layer.

Define each network path as an array of layer objects, and assign names to the input and output layers of each path. To create a recurrent neural network, use sequenceInputLayer as the input layer and include an lstmLayer as one of the other network layers.

% Define observation path.
obsPath = [
    sequenceInputLayer(prod(obsInfo.Dimension),Name="obsInLyr")
    fullyConnectedLayer(40)
    reluLayer
    fullyConnectedLayer(30,Name="obsOutLyr")
    ];

% Define action path.
actPath = [
    sequenceInputLayer(prod(actInfo.Dimension),Name="actInLyr")
    fullyConnectedLayer(30,Name="actOutLyr")
    ];

% Define common path.
commonPath = [
    concatenationLayer(1,2,Name="cat")
    lstmLayer(16)
    reluLayer
    fullyConnectedLayer(1)
    ];

% Create dlnetwork object and add layers.
criticNet = dlnetwork;
criticNet = addLayers(criticNet,obsPath);
criticNet = addLayers(criticNet,actPath);
criticNet = addLayers(criticNet,commonPath);

% Connect paths.
criticNet = connectLayers(criticNet,"obsOutLyr","cat/in1");
criticNet = connectLayers(criticNet,"actOutLyr","cat/in2");

To initialize the network weights differently for the two critics, create two different dlnetwork objects. You must do this because the agent constructor function does not accept two identical critics.

criticNet1 = initialize(criticNet);
criticNet2 = initialize(criticNet);

Display the number of weights.

summary(criticNet1)

   Initialized: true

   Number of learnables: 6.3k

   Inputs:
      1   'obsInLyr'   Sequence input with 2 dimensions
      2   'actInLyr'   Sequence input with 1 dimensions

Create the two critics using the two networks with different weights. Use the same network structure for both critics. The SAC agent initializes the two networks using different default parameters.

critic1 = rlQValueFunction(criticNet1,obsInfo,actInfo);
critic2 = rlQValueFunction(criticNet2,obsInfo,actInfo);

Check the critics with a random observation and action input.

getValue(critic1,{rand(obsInfo.Dimension)},{rand(actInfo.Dimension)})

ans = single

-0.0508

getValue(critic2,{rand(obsInfo.Dimension)},{rand(actInfo.Dimension)})

ans = single

0.0762

Since the critic has a recurrent network, the actor must have a recurrent network too.

Do not add a tanhLayer or scalingLayer in the mean output path. The SAC agent internally transforms the unbounded Gaussian distribution to the bounded distribution to compute the probability density function and entropy properly. However, add a softplus or ReLU layer to the standard deviations path to enforce nonnegativity,

Define each network path as an array of layer objects and specify a name for the input and output layers, so you can later explicitly associate them with the appropriate channel.

% Define common path.
commonPath = [
    sequenceInputLayer(prod(obsInfo.Dimension),Name="obsInLyr")
    fullyConnectedLayer(400)
    lstmLayer(8)
    reluLayer(Name="CommonOutLyr")
    ];

% Define mean value path.
meanPath = [
    fullyConnectedLayer(300,Name="MeanInLyr")
    reluLayer
    fullyConnectedLayer(prod(actInfo.Dimension),Name="MeanOutLyr")
    ];

% Define standard deviation value path.
stdPath = [
    fullyConnectedLayer(300,Name="StdInLyr")
    reluLayer
    fullyConnectedLayer(prod(actInfo.Dimension))
    softplusLayer(Name="StdOutLyr")
    ];

% Create dlnetwork object and add layers.
actorNet = dlnetwork;
actorNet = addLayers(actorNet,commonPath);
actorNet = addLayers(actorNet,meanPath);
actorNet = addLayers(actorNet,stdPath);

% Connect layers.
actorNet = connectLayers(actorNet,"CommonOutLyr","MeanInLyr/in");
actorNet = connectLayers(actorNet,"CommonOutLyr","StdInLyr/in");

% Initialize network and display the number of weights.
actorNet = initialize(actorNet);
summary(actorNet)

   Initialized: true

   Number of learnables: 20.2k

   Inputs:
      1   'obsInLyr'   Sequence input with 2 dimensions

Create the actor using actorNet, the observation and action specification objects, and the names of the input and output layers.

actor = rlContinuousGaussianActor(actorNet, obsInfo, actInfo, ...
    ActionMeanOutputNames="MeanOutLyr",...
    ActionStandardDeviationOutputNames="StdOutLyr",...
    ObservationInputNames="obsInLyr");

Check your actor with a random input observation.

getAction(actor,{rand(obsInfo.Dimension)})

ans = 1x1 cell array
    {[-0.6304]}

Specify training options for the critics. For more information, see rlOptimizerOptions.

criticOptions = rlOptimizerOptions( ...
    Optimizer = "adam", LearnRate = 1e-3,... 
    GradientThreshold = 1, L2RegularizationFactor = 2e-4);

Specify training options for the actor.

actorOptions = rlOptimizerOptions( ...
    Optimizer = "adam", LearnRate = 1e-3,...
    GradientThreshold = 1, L2RegularizationFactor = 1e-5);

Specify agent options. For more information, see rlSACAgentOptions.

agentOptions = rlSACAgentOptions;

To ensure that if you use the agent in an RL Agent block the block executes at the same sample time as the environment, set the agent sample time equal to the environment sample time. This also ensures that the time interval between consecutive elements in the returned experience is env.Ts.

agentOptions.SampleTime = env.Ts;

A large discount factor value of 0.99 promotes shorter term rewards.

agentOptions.DiscountFactor = 0.9;

Specify a larger smooth factor to promote a faster target critic update, possibly at the expense of noisier gradient estimate.

agentOptions.TargetSmoothFactor = 2e-3;

Specify a capacity of 1e6 for the experience buffer to store a diverse set of experiences.

agentOptions.ExperienceBufferLength = 1e6;

Use mini-batches of 256 experiences. Smaller mini-batches are computationally efficient but may introduce variance in training. By contrast, larger batch sizes can make the training stable but require higher memory.

agentOptions.MiniBatchSize = 32;

To use a recurrent neural network, you must specify a SequenceLength greater than 1.

agentOptions.SequenceLength = 32;

Include the actor and critic options in the agent options object.

agentOptions.CriticOptimizerOptions = criticOptions;
agentOptions.ActorOptimizerOptions = actorOptions;

Create the SAC agent using actor, critics, and option objects.

agent = rlSACAgent(actor,[critic1 critic2],agentOptions);

To check your agent, use getAction to return the action from a random observation.

getAction(agent,{rand(obsInfo.Dimension)})

ans = 1x1 cell array
    {[-0.8774]}

To evaluate the agent using sequential observations, use the sequence length (time) dimension. For example, obtain actions for a sequence of 9 observations.

[action,state] = getAction(agent, ...
    {rand([obsInfo.Dimension 1 9])});

Display the action corresponding to the seventh element of the observation.

action = action{1};
action(1,1,1,7)

ans = 
0.5114

You can now test and train the agent within the environment.

Version History

Introduced in R2020b