TABLE 2.

Dynamic parameters and their respective range of values used during training and testing. A larger range of values are used during testing to evaluate the generalization ability of control policies in dynamics uncertainties.

Dynamic parameters	Training range	Testing range
Friction coefficient	[0.9,1.6]*default value	[0.7,2.0]*default value
Mass	[0.8,1.2]*default value	[0.7,1.5]*default value
Motor strength	[0.8,1.2]*default value	[0.7,1.3]*default value
Observation latency	[0,0.04]s	[0,0.06]s
Inertial	[0.5,1.5]*default value	[0.4,1.6]*default value
Center of Mass	[0.9,1.2]*default value	[0.8,1.3]*default value