Implement contact models library #7

GiulioRomualdi · 2020-01-09T18:03:05Z

This PR introduces the contact models library.

In details:

An abstract interface named ContactModel has been implemented. This class will be useful to implement different contact models
The ContinuousContactModel implements the a continuous contact model.
C++17 is now required. (i.e. the library widly uses std::any)

- Implement the first version ContactModel class

…odels library

…ontinuousContactModel

…ontactModel

traversaro

Minor comments.

CMakeLists.txt

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContactModel.tpp

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContinuousContactModel.h

traversaro · 2020-01-13T07:56:40Z

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContinuousContactModel.h

+ */
+class ContinuousContactModel final : public ContactModel
+{
+    iDynTree::Transform m_frameTransform; /**< Homogeneous transformation of the frame placed in the


Of the frame "place in the center of the contact surface", w.r.t. to which other frame?

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContactModel.h

src/ContactModels/tests/ContinousContactModelTest.cpp

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContactModel.h

traversaro · 2020-01-13T08:44:27Z

Ah, a minor comment: it may be possible the unordered_set<string, any> structure for methods that are supposed to be called at each control loop may may be not efficient in term of computation time, so if you ever want to optimize the code this could be one of the things to check.

S-Dafarra · 2020-01-13T08:49:31Z

it may be possible the unordered_set<string, any> structure for methods that are supposed to be called at each control loop may may be not efficient in term of computation time

Is it? Search, insertion, and removal have average constant-time complexity (https://en.cppreference.com/w/cpp/container/unordered_map).

As far as I understood unordered_map is preferable to map since the latter orders the elements and performs a binary search, while the former stores the element in a hash table.

traversaro · 2020-01-13T08:53:44Z

it may be possible the unordered_set<string, any> structure for methods that are supposed to be called at each control loop may may be not efficient in term of computation time

Is it? Search, insertion, and removal have average constant-time complexity (https://en.cppreference.com/w/cpp/container/unordered_set)

Constant is not negligible, especially if any of those operation uses dynamic memory allocation. To use an hyperbole, If the constant time complexity is 100 ms, it would be indeed constant, but also unusable in practice. However I think is perfectly fine to ignore this for now and come back to it to see if it is actually relevant if we need to speed up the code.

S-Dafarra · 2020-01-13T08:58:58Z

Constant is not negligible, especially if any of those operation uses dynamic memory allocation.

As long as you don't add or remove elements, I don't think it does. There could be a difference when you need to scan all the elements given that they may be stored in non-contiguous portions of memory. In that case, vectors are always the best pick.

Once I had to find all the equal strings in a very large set of strings. A simple iterative algorithm was taking minutes. I then decided to use the strings as keys in an unordered map<string, vector<size_t>> and saved the corresponding index in the vector. At the end, it was enough to count the elements in each vector to know the number of equal strings. This was taking milliseconds instead.

…ontact_models

S-Dafarra · 2020-02-07T11:39:03Z

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContactModel.h

+     * @return true/false in case of success/failure
+     */
+    template <class T>
+    bool getVariable(const std::unordered_map<std::string, std::any>& variables,


I would detach VariablesHandler from the OptimalControlUtilities and use it also here.

I would do it later with another PR

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContinuousContactModel.h

S-Dafarra · 2020-02-07T13:08:03Z

src/ContactModels/src/ContinuousContactModel.cpp

+    // the parameters has been update the previous quantities has to be evaluated again
+    m_isContactWrenchComputed = false;
+    m_isControlMatrixComputed = false;
+    m_isAutonomusDynamicsComputed = false;


To be honest I don't like too much that child classes have to set a parameter that is used in the implementation of the base class. IMHO, the setting of these parameters to false should be handled by the base class or through a method in the base class, like invalidateBuffers() or similar.

I implemented the solution you suggested in #33 (comment)

Code: f4092bf

S-Dafarra · 2020-02-07T13:18:26Z

src/ContactModels/src/ContinuousContactModel.cpp

+    // compute the torque
+    auto skewRe1 = iDynTree::skew(rotation.col(0));
+    auto skewRe2 = iDynTree::skew(rotation.col(1));
+    torque = std::abs(rotation(2, 2)) * area / 12 * (m_length * m_length


Maybe it would be better to have a link to reference how this formula is obtained.

I can update the doxygen documentation adding the explanation of the model

src/ContactModels/tests/CMakeLists.txt

S-Dafarra · 2020-02-07T13:25:27Z

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContinuousContactModel.h

+ * as a continuum of springs and dampers.
+ * The contact surface is supposed to be <b>rectangular</b> and the frame placed on the contact
+ * surfaced is centered in the middle of the surface, with the \a z-axis pointing upwards and the \a
+ * x-axis pointing forward (parallel to one edge of the rectangle).


Maybe you can define here what do you mean by length and width. In addition, I would mention that this model assumes infinite friction.

Co-Authored-By: Stefano Dafarra <stefano.dafarra@gmail.com>

GiulioRomualdi · 2020-03-25T12:59:25Z

I updated the code accordingly to #28

I notice that we are implementing the initialize() method for each class and in general the class cannot be used till the initialize() function is called.
Probably we can generalize the concept by designing a class block and all the other classes will inherit from that class.

This is an example of the class

class Block
{
protected:
    enum class State
    {
         NotInitialized, Initialized, ...
    };
   State m_state;
public:
   virtual bool initialize(std::weak_ptr<IParameterHandler> handler) = 0;
};

We can discuss this on another issue

@S-Dafarra you can review again the code

DanielePucci · 2020-03-25T20:36:16Z

CC @gabrielenava

…ss anymore

…ontact_models

…dels

…tModels/include/BipedalLocomotion/* and improve the documentation

GiulioRomualdi · 2020-05-28T14:33:07Z

I updated the code to be compliant with the latest version of master. And I think I also handled all @S-Dafarra comments.

Let me know if we can merge this PR since it is blocking #46 😄

S-Dafarra · 2020-05-28T14:38:25Z

Go for me!

GiulioRomualdi added 9 commits January 8, 2020 16:56

- Implement the skeleton of the ContactModels library

b439002

- Implement the first version ContactModel class

Update the depenecies in order to compile the ContactModels library

3f00109

Enable the ContactModels library compilation

a73eb50

Introduce state and parameters in ContactModels library

99a193d

Require C++17 in the ContactModels library

8fc0884

Update the badge C++14 badge to C++17 in the README.md

136de41

Implement the first version of the ContinuousContactModel in ContactM…

a741933

…odels library

Add skeleton for testing in BipedalLocomotionControllers

524fc57

Implement the test for checking the Continuous contact model

c7b2df6

GiulioRomualdi self-assigned this Jan 9, 2020

GiulioRomualdi added 2 commits January 12, 2020 22:40

Implement computeControlMatrix() and computeAutonomousDynamics() in C…

600fce8

…ontinuousContactModel

Implement the test relared to the contact dynamics for the ContinousC…

3df976e

…ontactModel

GiulioRomualdi added this to the Iteration 35 - Walking awakens milestone Jan 12, 2020

GiulioRomualdi marked this pull request as ready for review January 12, 2020 22:00

GiulioRomualdi requested review from S-Dafarra and traversaro January 12, 2020 22:00

traversaro requested changes Jan 13, 2020

View reviewed changes

traversaro reviewed Jan 13, 2020

View reviewed changes

src/ContactModels/include/BipedalLocomotionControllers/ContactModels/ContactModel.h Outdated Show resolved Hide resolved

GiulioRomualdi added 2 commits January 17, 2020 12:21

Merge remote-tracking branch 'origin/master' into feature/implement_c…

cf5aae0

…ontact_models

Bugfix in ContinousContactModelTest.cpp

8da427d

GiulioRomualdi modified the milestones: Iteration 35 - Walking awakens, Iteration 36 - Release of the release Feb 4, 2020

GiulioRomualdi requested a review from MiladShafiee February 4, 2020 13:10

GiulioRomualdi added the enhancement New feature or request label Feb 4, 2020

S-Dafarra reviewed Feb 7, 2020

View reviewed changes

S-Dafarra mentioned this pull request Feb 7, 2020

Implement ParametersHandler library #13

Merged

GiulioRomualdi and others added 4 commits March 13, 2020 14:53

Improve documentation in ContinuousContactModel.h

db6e50c

Fix typo in ContatModel library

430db97

Fix documentation in ContactModel.h

089bce7

Fix typos in ContactModels library

c68d264

Co-Authored-By: Stefano Dafarra <stefano.dafarra@gmail.com>

GiulioRomualdi mentioned this pull request Mar 15, 2020

Decide how to proceed with the IParameterHandler interface #27

Closed

GiulioRomualdi added 3 commits March 25, 2020 11:24

Merge branch 'master' into feature/implement_contact_models

7044d40

Update the interface of the ContactModel library accordingly to 01d406c

7a64ab0

Update the ContactModels tests accordingly to 7a64ab0

604dd35

GiulioRomualdi modified the milestones: Iteration 39 - Working from home, Iteration 40 - Still at home Mar 25, 2020

The flags defined in ContactModel class are not used in the child cla…

f4092bf

…ss anymore

GiulioRomualdi modified the milestones: Iteration 40 - Still at home, Iteration 41 - Easter from home Apr 8, 2020

GiulioRomualdi added 6 commits May 28, 2020 15:53

Merge remote-tracking branch 'origin/master' into feature/implement_c…

fb4151b

…ontact_models

Renaming includes and cmake functions accordingly to fb4151b

5e96e04

Implement the regressor in ContactModel

49d836d

Detach the nullForceTransform from the setState function in ContactMo…

8ad69b0

…dels

Update the test accordingly to the latest modifications

0b1739d

Move ContactModels/include/BipedalLocomotionControllers/* into Contac…

7043c20

…tModels/include/BipedalLocomotion/* and improve the documentation

GiulioRomualdi force-pushed the feature/implement_contact_models branch from 50227ea to 7043c20 Compare May 28, 2020 14:23

Fix typo in BipedalLocomotionFrameworkFindDependencies.cmake

ea80f30

GiulioRomualdi added this to the Iteration 44 - Like the cats milestone May 28, 2020

traversaro approved these changes May 28, 2020

View reviewed changes

GiulioRomualdi merged commit ec1c7b1 into ami-iit:master May 28, 2020

GiulioRomualdi deleted the feature/implement_contact_models branch May 28, 2020 14:52

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Implement contact models library #7

Implement contact models library #7

GiulioRomualdi commented Jan 9, 2020 •

edited

traversaro left a comment

traversaro Jan 13, 2020

traversaro commented Jan 13, 2020

S-Dafarra commented Jan 13, 2020 •

edited

traversaro commented Jan 13, 2020

S-Dafarra commented Jan 13, 2020 •

edited

S-Dafarra Feb 7, 2020

GiulioRomualdi May 28, 2020

S-Dafarra Feb 7, 2020

GiulioRomualdi Mar 27, 2020

GiulioRomualdi Mar 27, 2020 •

edited

S-Dafarra Feb 7, 2020

GiulioRomualdi Mar 13, 2020

S-Dafarra Feb 7, 2020

GiulioRomualdi commented Mar 25, 2020

DanielePucci commented Mar 25, 2020

GiulioRomualdi commented May 28, 2020

S-Dafarra commented May 28, 2020

Implement contact models library #7

Implement contact models library #7

Conversation

GiulioRomualdi commented Jan 9, 2020 • edited

traversaro left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

traversaro commented Jan 13, 2020

S-Dafarra commented Jan 13, 2020 • edited

traversaro commented Jan 13, 2020

S-Dafarra commented Jan 13, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

GiulioRomualdi Mar 27, 2020 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

GiulioRomualdi commented Mar 25, 2020

DanielePucci commented Mar 25, 2020

GiulioRomualdi commented May 28, 2020

S-Dafarra commented May 28, 2020

GiulioRomualdi commented Jan 9, 2020 •

edited

S-Dafarra commented Jan 13, 2020 •

edited

S-Dafarra commented Jan 13, 2020 •

edited

GiulioRomualdi Mar 27, 2020 •

edited