dm-robotics-manipulation 0.0.3

Parametrically defined mesh objects

Objects for robotic manipulation.

This folder provides various physical objects and software utilities used to build simulated environments for robotic manipulation. In particular, it provides a way to generate parametric objects with OnShape's Python API.

An object in simulation.

It's worth pointing out that in the context of this codebase when we refer to "objects" we don't have in mind a software object but rather a physical object with attributes such as mass, shape and texture. The codebase helps with generating objects representations (e.g. .STL files) with the OnShape python API client. OnShape is cloud-based CAD software which helps creating three-dimensional shapes and their parameterizations.

Parametric objects

An important component of this codebase is the ParametricObject class which can be used to represent a parametric object and check its consitency. A ParametricObject is a collection of two ParametricProperties, one representing the shape and the other representing the texture.

Parametric objects documentation

Parametric RGB-objects 🛑🟩🔷

An example of ParametricObject are the RGB-objects. The RGB-objects are a set of parametric objects designed in OnShape. Their shape has been designed to offer a multitude of different affordances for a parallel gripper grapsing and stacking. The basic idea behind RGB-objects' shapes is to start from a 2D shape and to extrude this shape along an axis. Both the 2D shape and the extrusion are parameterized by a few parameters, described in the documentation page.

RGB-objects 🛑🟩🔷 documentation

Usage example

RGB-objects are located in props/rgb_objects and can be initialized as following.

from dm_robotics.manipulation.props.rgb_objects import rgb_object

 color_set = [
      [1, 0, 0, 1],  # RED
      [0, 1, 0, 1],  # GREEN
      [0, 0, 1, 1],  # BLUE
  ]

# Only a name of the object and its color are required. Optional parameters
# include scaling, mass etc.
prop_1 = rgb_object.RgbObjectProp(obj_id='r3', color=color_set[0])
prop_2 = rgb_object.RgbObjectProp(obj_id='g2', color=color_set[1])
prop_3 = rgb_object.RgbObjectProp(obj_id='b1', color=color_set[2])

# When a list of all available objects could be accessed with
ids_list =  rgb_object.ids_list

Objects could be created individually as in the above snippet. Some predefined object sets have been defined in rgb_object.PROP_SETS. These sets can be used for training and evaluation on the RGB-stacking task.

object_set = 'rgb_test_set1'
_, obj_ids = rgb_object.PROP_SETS[object_set]
task_props = []
for i, obj_id in enumerate(objects):
  prop = rgb_object.RgbObjectProp(
        obj_id=obj_id, color=color_set[i], name=f'rgb_object_{obj_id}')
  task_props.append(prop)

We also provide some utilities to randomly sample objects. The following snippet allows to create sets of random objects at every call of the function PROP_SETS[object_set].

object_set = 'rgb_train_random'

for i in range(num_episodes):
  _, obj_ids = rgb_object.PROP_SETS[object_set]
  # `obj_ids` will provide different object ids for every access.