Unity 机器学习代理工具包 (ML-Agents) 是一个开源项目,它使游戏和模拟能够作为训练智能代理的环境。
您最多选择25个主题 主题必须以中文或者字母或数字开头,可以包含连字符 (-),并且长度不得超过35个字符
 
 
 
 
 

115 行
4.0 KiB

using System;
using UnityEngine;
using UnityEngine.Serialization;
namespace Unity.MLAgents.Policies
{
/// <summary>
/// Whether the action space is discrete or continuous.
/// </summary>
public enum SpaceType
{
/// <summary>
/// Discrete action space: a fixed number of options are available.
/// </summary>
Discrete,
/// <summary>
/// Continuous action space: each action can take on a float value.
/// </summary>
Continuous
}
/// <summary>
/// Holds information about the brain. It defines what are the inputs and outputs of the
/// decision process.
/// </summary>
/// <remarks>
/// Set brain parameters for an <see cref="Agent"/> instance using the
/// <seealso cref="BehaviorParameters"/> component attached to the agent's [GameObject].
///
/// [GameObject]: https://docs.unity3d.com/Manual/GameObjects.html
/// </remarks>
[Serializable]
public class BrainParameters
{
/// <summary>
/// The size of the observation space.
/// </summary>
/// <remarks>An agent creates the observation vector in its
/// <see cref="Agent.CollectObservations(Sensors.VectorSensor)"/>
/// implementation.</remarks>
/// <value>
/// The length of the vector containing observation values.
/// </value>
[FormerlySerializedAs("vectorObservationSize")]
public int VectorObservationSize = 1;
/// <summary>
/// Stacking refers to concatenating the observations across multiple frames. This field
/// indicates the number of frames to concatenate across.
/// </summary>
[FormerlySerializedAs("numStackedVectorObservations")]
[Range(1, 50)] public int NumStackedVectorObservations = 1;
/// <summary>
/// The size of the action space.
/// </summary>
/// <remarks>The size specified is interpreted differently depending on whether
/// the agent uses the continuous or the discrete action space.</remarks>
/// <value>
/// For the continuous action space: the length of the float vector that represents
/// the action.
/// For the discrete action space: the number of branches in the action space.
/// </value>
[FormerlySerializedAs("vectorActionSize")]
public int[] VectorActionSize = new[] {1};
/// <summary>
/// The list of strings describing what the actions correspond to.
/// </summary>
[FormerlySerializedAs("vectorActionDescriptions")]
public string[] VectorActionDescriptions;
/// <summary>
/// Defines if the action is discrete or continuous.
/// </summary>
[FormerlySerializedAs("vectorActionSpaceType")]
public SpaceType VectorActionSpaceType = SpaceType.Discrete;
/// <summary>
/// The number of actions specified by this Brain.
/// </summary>
public int NumActions
{
get
{
switch (VectorActionSpaceType)
{
case SpaceType.Discrete:
return VectorActionSize.Length;
case SpaceType.Continuous:
return VectorActionSize[0];
default:
return 0;
}
}
}
/// <summary>
/// Deep clones the BrainParameter object.
/// </summary>
/// <returns> A new BrainParameter object with the same values as the original.</returns>
public BrainParameters Clone()
{
return new BrainParameters
{
VectorObservationSize = VectorObservationSize,
NumStackedVectorObservations = NumStackedVectorObservations,
VectorActionSize = (int[])VectorActionSize.Clone(),
VectorActionDescriptions = (string[])VectorActionDescriptions.Clone(),
VectorActionSpaceType = VectorActionSpaceType
};
}
}
}