diff --git a/pyrobolearn/optimizers/__init__.py b/pyrobolearn/optimizers/__init__.py
index 9168742..fc92f54 100644
--- a/pyrobolearn/optimizers/__init__.py
+++ b/pyrobolearn/optimizers/__init__.py
@@ -1,3 +1,27 @@
 
-# import optimizers
-from optimizer import *
+# import optimizer
+from optimizer import Optimizer
+
+# import scipy optimizer
+# from scipy_optimizer import Scipy
+
+# import nlopt optimizer
+# from nlopt_optimizer import NLopt
+
+# import ipopt optimizer
+# from ipopt_optimizer import IPopt
+
+# import QP solvers
+# from qpsolvers_optimizer import QP
+
+# import CMA-ES optimizer
+# from cma_optimizer import CMAES
+
+# import Bayesian Optimizer
+# from gpyopt_optimizer import BayesianOptimizer
+
+# import torch optimizers
+from torch_optimizer import *
+
+# import Contact-Invariant Optimizer
+# from cio import CIO
diff --git a/pyrobolearn/optimizers/cio.py b/pyrobolearn/optimizers/cio.py
index 65b78c5..ab28880 100644
--- a/pyrobolearn/optimizers/cio.py
+++ b/pyrobolearn/optimizers/cio.py
@@ -1,6 +1,12 @@
-# This file implements the 'Contact Invariant Optimization' framework developed by Igor Mordatch.
-# Ref: "Automated Discovery and Learning of Complex Movement Behaviors" (PhD thesis), Mordatch, 2015
-# See also: presentation given CS294
+#!/usr/bin/env python
+"""This file implements the 'Contact Invariant Optimization' framework developed by Igor Mordatch.
+
+References:
+    [1] "Automated Discovery and Learning of Complex Movement Behaviors" (PhD thesis), Mordatch, 2015
+    [2] Mordatch's presentation given in CS294
+"""
+# TODO: this is not an optimizer, but more an optimization process. It should be in another directory, maybe in
+#  `trajectory_optimization`??
 
 import numpy as np
 from scipy.interpolate as interp1d
@@ -56,8 +62,8 @@ class CIO(object):
         return self.phase(t)
 
     def compute_state(self):
-        base_pos = self.robot.getBasePosition()
-        base_quat = self.robot.getBaseOrientation()
+        base_pos = self.robot.get_base_position()
+        base_quat = self.robot.get_base_orientation()
         end_effector_pos = self.robot.getEndEffectorPositions()
         end_effector_quat = self.robot.getEndEffectorOrientations()
 
diff --git a/pyrobolearn/optimizers/torch_optimizer.py b/pyrobolearn/optimizers/torch_optimizer.py
index a387475..1d86582 100644
--- a/pyrobolearn/optimizers/torch_optimizer.py
+++ b/pyrobolearn/optimizers/torch_optimizer.py
@@ -39,7 +39,7 @@ __status__ = "Development"
 #         pass
 
 
-class Adam(object):
+class Adam(Optimizer):
     r"""Adam Optimizer
 
     References:
@@ -47,7 +47,8 @@ class Adam(object):
     """
 
     def __init__(self, learning_rate=1e-3, betas=(0.9, 0.999), eps=1e-08, weight_decay=0, amsgrad=False,
-                 max_grad_norm=None):  # 0.5
+                 max_grad_norm=None, *args, **kwargs):  # 0.5
+        super(Adam, self).__init__(*args, **kwargs)
         self.optimizer = None
         self.learning_rate = learning_rate
         self.betas = betas
@@ -73,14 +74,15 @@ class Adam(object):
         self.optimizer.step()
 
 
-class Adadelta(object):
+class Adadelta(Optimizer):
     r"""Adadelta Optimizer
 
     References:
         [1] "ADADELTA: An Adaptive Learning Rate Method", Zeiler, 2012
     """
 
-    def __init__(self, learning_rate=1., rho=0.9, eps=1e-6, weight_decay=0, max_grad_norm=None): #0.5
+    def __init__(self, learning_rate=1., rho=0.9, eps=1e-6, weight_decay=0, max_grad_norm=None, *args, **kwargs):  # 0.5
+        super(Adadelta, self).__init__(*args, **kwargs)
         self.optimizer = None
         self.learning_rate = learning_rate
         self.rho = rho
@@ -101,7 +103,7 @@ class Adadelta(object):
         self.optimizer.step()
 
 
-class Adagrad(object):
+class Adagrad(Optimizer):
     r"""Adagrad Optimizer
 
     References:
@@ -109,7 +111,8 @@ class Adagrad(object):
     """
 
     def __init__(self, learning_rate=0.01, learning_rate_decay=0, weight_decay=0, initial_accumumaltor_value=0,
-                 max_grad_norm=None):  # 0.5
+                 max_grad_norm=None, *args, **kwargs):  # 0.5
+        super(Adagrad, self).__init__(*args, **kwargs)
         self.optimizer = None
         self.learning_rate = learning_rate
         self.learning_rate_decay = learning_rate_decay
@@ -131,7 +134,7 @@ class Adagrad(object):
         self.optimizer.step()
 
 
-class RMSprop(object):
+class RMSprop(Optimizer):
     r"""RMSprop
 
     References:
@@ -141,7 +144,8 @@ class RMSprop(object):
     """
 
     def __init__(self, learning_rate=1e-2, alpha=0.99, eps=1e-8, weight_decay=0, momentum=0, centered=False,
-                 max_grad_norm=None):  # 0.5
+                 max_grad_norm=None, *args, **kwargs):  # 0.5
+        super(RMSprop, self).__init__(*args, **kwargs)
         self.optimizer = None
         self.learning_rate = learning_rate
         self.alpha = alpha
@@ -165,7 +169,7 @@ class RMSprop(object):
         self.optimizer.step()
 
 
-class SGD(object):
+class SGD(Optimizer):
     r"""Stochastic Gradient Descent
 
     References:
@@ -173,8 +177,9 @@ class SGD(object):
         [2] "On the importance of initialization and momentum in deep learning", Sutskever et al., 2013
     """
 
-    def __init__(self, learning_rate=1e-3, momentum=0, dampening=0, weight_decay=0, nesterov=False,
-                 max_grad_norm=None): #0.5
+    def __init__(self, learning_rate=1e-3, momentum=0, dampening=0, weight_decay=0, nesterov=False, max_grad_norm=None,
+                 *args, **kwargs):  # 0.5
+        super(SGD, self).__init__(*args, **kwargs)
         self.optimizer = None
         self.learning_rate = learning_rate
         self.momentum = momentum