BipedalLocomotion::ML::MANN class
#include <BipedalLocomotion/ML/MANN.h>

MANN is a class that allows to perform the inference of an onnx implementing Mode-Adaptive Neural Networks (MANN).

The network was originally proposed in [H. Zhang, S. Starke, T. Komura, and J. Saito, “Mode-adaptive neural networks for quadruped motion control,” ACM Trans. Graph., vol. 37, no. 4, pp. 1–11, 2018.](https://doi.org/10.1145/3197517.3201366) This class uses onnxruntime to perform inference and it has been tested with an onnx model generated from https://github.com/ami-iit/mann-pytorch

Base classes

template<class _Input, class _Output> class BipedalLocomotion::System::Advanceable<MANNInput, MANNOutput>: Basic class that represents a discrete system.

Constructors, destructors, conversion operators

MANN(): Constructor.
~MANN(): Destructor.

Public functions

auto initialize(std::weak_ptr<const ParametersHandler::IParametersHandler> paramHandler) -> bool override: Initialize the network.
auto setInput(const MANNInput& input) -> bool override: Set the input of the network.
auto advance() -> bool override: Perform one step of the inference given the input set.
auto getOutput() const -> const MANNOutput& override: Get the output of the network once advance has been called.
auto isOutputValid() const -> bool override: Check if the output of the network is valid.

Function documentation

BipedalLocomotion::ML::MANN::~MANN()

Destructor.

It is required by the pimpl implementation.

bool BipedalLocomotion::ML::MANN::initialize(std::weak_ptr<const ParametersHandler::IParametersHandler> paramHandler) override

Initialize the network.

Parameters
paramHandler	pointer to the parameters handler.
Returns	True in case of success, false otherwise.

Note

the following parameters are required by the class

Parameter Name	Type	Description	Mandatory
`onnx_model_path`	`string`	Path to the `onnx` model that will be loaded to perform inference	Yes
`number_of_joints`	`int`	Number of robot joints considered in the model	Yes
`projected_base_datapoints`	`int`	Number of samples of the base horizon considered in the model (it must be an even number)	Yes

bool BipedalLocomotion::ML::MANN::setInput(const MANNInput& input) override

Set the input of the network.

Parameters
input	the struct containing all the inputs of the network.
Returns	true if case of success, false otherwise.

bool BipedalLocomotion::ML::MANN::advance() override

Perform one step of the inference given the input set.

Returns	true if case of success, false otherwise.

const MANNOutput& BipedalLocomotion::ML::MANN::getOutput() const override

Get the output of the network once advance has been called.

Returns	the output of the network.

bool BipedalLocomotion::ML::MANN::isOutputValid() const override

Check if the output of the network is valid.

Returns	true if it is valid, false otherwise.