ROS进二阶学习笔记（2）- SMACH：用状态机来管理机器人任务
ROS进二阶学习笔记（2） - SMACH：用状态机来管理机器人任务
http://wiki.ros.org/smach/Tutorials/Getting Started
http://wiki.ros.org/smach/Tutorials
好久没有更新关于ROS的笔记，这里进入SMACH的学习和应用，参照rbx2(ros by example volume 2)，和ROS官方wiki，进行学习。
====
关于概念
SMACH states是Python类，通过重写execute()方法来返回一个或多个可能的outcomes, 从而扩展smach.State类。 excute() 方法也可以通过可选的参数来来定义一个userdata集合，这个集合可以被用来传递状态之间的信息。本质上讲，状态执行的实际计算基本上可以是任何你想要的。并且，有很多预先定义好的state types ，这可以节省大量的非必须的代码。
- A SMACH state machine is another Python class (smach.StateMachine) that can contain a number of states.
- A state is added to a state machine by defining a set of transitions from the state’s outcomes to other states in the machine.
When a state machine is run, these transitions determine the flow of execution from state to state:
input → STATE_1 → {outcome, transition} → STATE_2
input → STATE_2 → {outcome, transition} → STATE_3
etc.
A state machine itself must have an outcome and can therefore be a state in another state machine. In this way, state machines can be nested to form a hierarchy. For example, a state machine called “clean house” could contain the nested state machines “vacuum living room “, “mop kitchen”, “wash tub” and so on. The higher level state machine “clean house” will determine how the outcomes of the nested state machines map into the overall outcome (e.g. “all tasks complete” or “not all tasks complete”).
预先实现的状态库
Smach comes with a whole library of pre-implemented states that cover many common usecases:
- 特别是SimpleActionState 类，把一个普通的ROS action 变成了一个SMACH 状态。
- 类似的，MonitorState 类封装了ROS topic;
- ServiceState类处理ROS Services;
- CBState 使用 @smach.cb_interface 装饰器使得你可以将几乎任何函数都变成一个SMACH 状态。
Container
To create a Smach state machine, you first create a number of states, and then add those states to a State Machine container.
There are a number of predefined SMACH containers that can also save you a lot of programming.
- The Concurrence 协作 container returns an outcome that depends on more than one state and allows one state to preempt抢占 another.
- The Sequence 序列 container automatically generates sequential transitions among the states that are added to it.
- The Iterator 迭代 container allows you to loop through one or more states until some condition is met.
We will learn more about these containers as we need them.