Appending visual and collision models

[florent-lamiraux]

pinocchio/include/pinocchio/algorithm/model.hpp

Line 77 in 8f65273

template<typename Scalar, int Options, template<typename, int> class JointCollectionTpl>

In the above method, a pair (model, geometric model) is appended to an existing pair of the same type.
It seems that this method is mainly used in HPP software where only the collision model is handled.
Appending a triple (model, collision model, visual model) to an existing triple of the same type is therefore not trivial.

Do you think relevant to add a method with prototype:

  template<typename Scalar, int Options, template<typename, int> class JointCollectionTpl>
  void appendModel(
    const ModelTpl<Scalar, Options, JointCollectionTpl> & modelA,
    const ModelTpl<Scalar, Options, JointCollectionTpl> & modelB,
    const GeometryModel & collisionModelA,
    const GeometryModel & collisionModelB,
    const GeometryModel & visualModelA,
    const GeometryModel & visualModelB,
    const FrameIndex frameInModelA,
    const SE3Tpl<Scalar, Options> & aMb,
    ModelTpl<Scalar, Options, JointCollectionTpl> & model,
    GeometryModel & collisionModel,
    GeometryModel & visualModel);

that would take care of both visual and collision models ?

[jcarpent]

I would say calling twice the function is sufficient. But in parallel, we have done at Willow a prototype  which takes as input a std::vector of GeometryModels.

…

-----Message original----- De: Florent ***@***.***> à: stack-of-tasks ***@***.***> Cc: Subscribed ***@***.***> Envoyé: mercredi 3 septembre 2025 16:10 CEST Sujet : [stack-of-tasks/pinocchio] Appending visual and collision models (Discussion #2752)   https://github.com/stack-of-tasks/pinocchio/blob/8f65273da9d7a48e210bf4797eda2e6e3259a0a6/include/pinocchio/algorithm/model.hpp#L77 In the above method, a pair (model, geometric model) is appended to an existing pair of the same type. It seems that this method is mainly used in HPP software where only the collision model is handled. Appending a triple (model, collision model, visual model) to an existing triple of the same type is therefore not trivial. Do you think relevant to add a method with prototype: template<typename Scalar, int Options, template<typename, int> class JointCollectionTpl> void appendModel( const ModelTpl<Scalar, Options, JointCollectionTpl> & modelA, const ModelTpl<Scalar, Options, JointCollectionTpl> & modelB, const GeometryModel & collisionModelA, const GeometryModel & collisionModelB, const GeometryModel & visualModelA, const GeometryModel & visualModelB, const FrameIndex frameInModelA, const SE3Tpl<Scalar, Options> & aMb, ModelTpl<Scalar, Options, JointCollectionTpl> & model, GeometryModel & collisionModel, GeometryModel & visualModel); that would take care of both visual and collision models ? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[florent-lamiraux]

Indeed, I will try.