# mujoco environment in openai gym: observation and action explanation and control

I am new to RL and mujoco. I just set up mujoco and am testing the FetchPickAndPlace environment. I called the following methods:

env.action_space.sample()


returns array([ 0.14159757, -0.6169547 , 0.56277525, 0.21222822], dtype=float32) and env.reset() returns

    {'observation': array([ 1.34782473e+00,  7.48951529e-01,  4.13637614e-01,  1.42999837e+00,
6.54897670e-01,  4.24702091e-01,  8.21736315e-02, -9.40538593e-02,
1.10644764e-02, -1.50186348e-06,  1.24443593e-03, -3.85214084e-07,
5.92637053e-07,  1.13569317e-13,  8.95029727e-04, -4.10490913e-05,
4.90967749e-05,  1.88857148e-07, -2.90549459e-07,  4.01961762e-18,
-8.95036946e-04,  4.10443989e-05, -2.26900915e-06,  3.77638244e-07,
4.38184449e-05]),
'achieved_goal': array([1.42999837, 0.65489767, 0.42470209]),
'desired_goal': array([1.48356396, 0.62833857, 0.42469975])}


Could someone explain the meanings of the numbers in the arrays?

Control: is it possible to manually write a program to manipulate the robot to achieve the goal (grab the object first, then move to the goal).