Pause Functionality#28
Conversation
sangteak601
left a comment
sangteak601
left a comment
There was a problem hiding this comment.
Hi @mrceki,
I'm still in the process of reviewing, but I have a quick question to help me understand it better. Please bear with me.
| if (pause_simulation_) { | ||
| // when paused, use the wall clock | ||
| sim_time_ros = node_->get_clock()->now(); | ||
| sim_period = control_period_; | ||
| } | ||
| else { | ||
| // Get the simulation time and period | ||
| auto sim_time = mj_data_->time; | ||
| int sim_time_sec = static_cast<int>(sim_time); | ||
| int sim_time_nanosec = static_cast<int>((sim_time - sim_time_sec)*1000000000); | ||
|
|
||
| sim_time_ros = rclcpp::Time(sim_time_sec, sim_time_nanosec, RCL_ROS_TIME); | ||
| sim_period = sim_time_ros - last_update_sim_time_ros_; | ||
| } |
There was a problem hiding this comment.
Why do we need to separate these two cases?
I think it might cause a problem, particularly when we resume the simulation, as time will jump (and also go backward).
There was a problem hiding this comment.
I separated the cases to keep controller updates running even when simulation time is frozen. But I'm not completely sure if simulation time really stops when paused. I'll need to double-check that.
There was a problem hiding this comment.
This version does not lead to time jumping as I observed. But it is also unnecessary to separate it in two cases because it's working unlike what I remember about time freezing.
There was a problem hiding this comment.
@sangteak601 I think I found out why I separated in two cases. If you pause the sim without pausing at startup, that's OK. But If we want to pause simulation in startup, controllers cannot be loaded. What is your suggestion on that?
There was a problem hiding this comment.
@sangteak601 I think I found out why I separated in two cases. If you pause the sim without pausing at startup, that's OK. But If we want to pause simulation in startup, controllers cannot be loaded. What is your suggestion on that?
I have done testing and was able to reproduce the issue of controllers not being loaded. One potential solution is to move controller_manager_->update() outside the if (sim_period >= control_period_) condition, similar to how controller_manager_->write() is currently positioned.
However, I am uncertain whether this change might introduce performance overhead, and I am also unsure why I initially placed controller_manager_->write() outside the condition.
2 participants
Hi!
As discussed in issue #25, this pull request adds pause functionality. You can toggle the pause state via the space bar.
Also, you can trigger pause at startup using the
'pause_simulation': Trueparameter in the launch file.