Hi everyone,

I am like many others trying to train a reinforcement learning algorithm using gazebo to optimize PID gains. Using wrong gains can easily crash the control and make my robotic arm go swinging. To account for this I am implementing the reset simulation after each movement using:

reset_simulation = rospy.ServiceProxy('/gazebo/reset_simulation', Empty) 
reset_simulation()

Now after the reset no new actions can be executed until after the time of the previous action has passed. I have no idea how to clear these "old" goals from the memory. Furthermore I am not sure if this issue is general or due to special actuators of my project.

Thanks for your time in advance!