Optimal control problem solver by Riccati recursion. More...

#include <ocp_solver.hpp>

Public Member Functions
	OCPSolver (const OCP &ocp, const SolverOptions &solver_options=SolverOptions())
	Construct optimal control problem solver. More...

	OCPSolver ()
	Default constructor. More...

	~OCPSolver ()=default
	Default destructor. More...

	OCPSolver (const OCPSolver &)=default
	Default copy constructor. More...

OCPSolver &	operator= (const OCPSolver &)=default
	Default copy assign operator. More...

	OCPSolver (OCPSolver &&) noexcept=default
	Default move constructor. More...

OCPSolver &	operator= (OCPSolver &&) noexcept=default
	Default move assign operator. More...

void	setSolverOptions (const SolverOptions &solver_options)
	Sets the solver option. More...

void	discretize (const double t)
	Discretizes the problem and reiszes the data structures. More...

void	initConstraints ()
	Initializes the priaml-dual interior point method for inequality constraints. More...

void	solve (const double t, const Eigen::VectorXd &q, const Eigen::VectorXd &v, const bool init_solver=true)
	Solves the optimal control problem. Internally calls updateSolutio() and discretize(). More...

const SolverStatistics &	getSolverStatistics () const
	Gets the solver statistics. More...

const Solution &	getSolution () const
	Get the solution over the horizon. More...

const SplitSolution &	getSolution (const int stage) const
	Get the split solution of a time stage. For example, the control input torques at the initial stage can be obtained by ocp.getSolution(0).u. More...

std::vector< Eigen::VectorXd >	getSolution (const std::string &name, const std::string &option="") const
	Get the solution vector over the horizon. More...

const aligned_vector< LQRPolicy > &	getLQRPolicy () const
	Gets of the local LQR policies over the horizon. More...

const RiccatiFactorization &	getRiccatiFactorization () const
	Gets the Riccati factorizations. This can be interpreted as locally approximated cost-to-go functions. More...

void	setSolution (const Solution &s)
	Sets the solution guess over the horizon. More...

void	setSolution (const std::string &name, const Eigen::VectorXd &value)
	Sets the solution guess over the horizon. More...

double	KKTError (const double t, const Eigen::VectorXd &q, const Eigen::VectorXd &v)
	Computes the KKT residual of the optimal control problem and returns the KKT error, that is, the l2-norm of the KKT residual. More...

double	KKTError () const
	Returns the l2-norm of the KKT residuals using the results of OCPsolver::updateSolution() or OCPsolver::solve(). More...

const TimeDiscretization &	getTimeDiscretization () const
	Gets the discretization. More...

void	setRobotProperties (const RobotProperties &properties)
	Sets a collection of the properties for robot model in this solver. More...

void	disp (std::ostream &os) const
	Displays the optimal control problem solver onto a ostream. More...

Friends
std::ostream &	operator<< (std::ostream &os, const OCPSolver &ocp_solver)

Detailed Description

Optimal control problem solver by Riccati recursion.

Constructor & Destructor Documentation

◆ OCPSolver() [1/4]

robotoc::OCPSolver::OCPSolver	(	const OCP &	ocp,
		const SolverOptions &	solver_options = `SolverOptions()`
	)

Construct optimal control problem solver.

Parameters

[in]	ocp	Optimal control problem.
[in]	solver_options	Solver options. Default is SolverOptions().

◆ OCPSolver() [2/4]

robotoc::OCPSolver::OCPSolver ( )

Default constructor.

◆ ~OCPSolver()

robotoc::OCPSolver::~OCPSolver ( )

default

Default destructor.

◆ OCPSolver() [3/4]

robotoc::OCPSolver::OCPSolver ( const OCPSolver & )

default

Default copy constructor.

◆ OCPSolver() [4/4]

robotoc::OCPSolver::OCPSolver ( OCPSolver && )

defaultnoexcept

Default move constructor.

Member Function Documentation

◆ discretize()

void robotoc::OCPSolver::discretize ( const double t )

Discretizes the problem and reiszes the data structures.

Parameters

[in] t Initial time of the horizon.

◆ disp()

void robotoc::OCPSolver::disp ( std::ostream & os ) const

Displays the optimal control problem solver onto a ostream.

◆ getLQRPolicy()

const aligned_vector< LQRPolicy > & robotoc::OCPSolver::getLQRPolicy ( ) const

Gets of the local LQR policies over the horizon.

Returns: const reference to the local LQR policies.

◆ getRiccatiFactorization()

const RiccatiFactorization & robotoc::OCPSolver::getRiccatiFactorization ( ) const

Gets the Riccati factorizations. This can be interpreted as locally approximated cost-to-go functions.

Returns: const reference to the Riccati factorizations.

◆ getSolution() [1/3]

const Solution & robotoc::OCPSolver::getSolution ( ) const

Get the solution over the horizon.

Returns: const reference to the solution.

◆ getSolution() [2/3]

const SplitSolution & robotoc::OCPSolver::getSolution ( const int stage ) const

Get the split solution of a time stage. For example, the control input torques at the initial stage can be obtained by ocp.getSolution(0).u.

Parameters

[in] stage Time stage of interest. Must be larger than 0 and smaller than N.

Returns: Const reference to the split solution of the specified time stage.

◆ getSolution() [3/3]

std::vector< Eigen::VectorXd > robotoc::OCPSolver::getSolution	(	const std::string &	name,
		const std::string &	option = `""`
	)		const

Get the solution vector over the horizon.

Parameters

[in]	name	Name of the variable.
[in]	option	Option for the solution. If name == "f" and option == "WORLD", the contact forces expressed in the world frame is returned. if option is set to other values, these expressed in the local frame are returned.

Returns: Solution vector.

◆ getSolverStatistics()

const SolverStatistics & robotoc::OCPSolver::getSolverStatistics ( ) const

Gets the solver statistics.

Returns: Solver statistics.

◆ getTimeDiscretization()

const TimeDiscretization & robotoc::OCPSolver::getTimeDiscretization ( ) const

Gets the discretization.

Returns: Returns const reference to the time discretization.

◆ initConstraints()

void robotoc::OCPSolver::initConstraints ( )

Initializes the priaml-dual interior point method for inequality constraints.

◆ KKTError() [1/2]

double robotoc::OCPSolver::KKTError ( ) const

Returns the l2-norm of the KKT residuals using the results of OCPsolver::updateSolution() or OCPsolver::solve().

Returns: The l2-norm of the KKT residual.

◆ KKTError() [2/2]

double robotoc::OCPSolver::KKTError	(	const double	t,
		const Eigen::VectorXd &	q,
		const Eigen::VectorXd &	v
	)

Computes the KKT residual of the optimal control problem and returns the KKT error, that is, the l2-norm of the KKT residual.

Parameters

[in]	t	Initial time of the horizon.
[in]	q	Initial configuration. Size must be Robot::dimq().
[in]	v	Initial velocity. Size must be Robot::dimv().

Returns: The KKT error, that is, the l2-norm of the KKT residual.

Remarks: The linear and angular velocities of the floating base are assumed to be expressed in the body local coordinate.

◆ operator=() [1/2]

OCPSolver & robotoc::OCPSolver::operator= ( const OCPSolver & )

default

Default copy assign operator.

◆ operator=() [2/2]

OCPSolver & robotoc::OCPSolver::operator= ( OCPSolver && )

defaultnoexcept

Default move assign operator.

◆ setRobotProperties()

void robotoc::OCPSolver::setRobotProperties ( const RobotProperties & properties )

Sets a collection of the properties for robot model in this solver.

Parameters

[in] properties A collection of the properties for the robot model.

◆ setSolution() [1/2]

void robotoc::OCPSolver::setSolution ( const Solution & s )

Sets the solution guess over the horizon.

Parameters

[in] s Solution.

◆ setSolution() [2/2]

void robotoc::OCPSolver::setSolution	(	const std::string &	name,
		const Eigen::VectorXd &	value
	)

Sets the solution guess over the horizon.

Parameters

[in]	name	Name of the variable.
[in]	value	Value of the specified variable.

◆ setSolverOptions()

void robotoc::OCPSolver::setSolverOptions ( const SolverOptions & solver_options )

Sets the solver option.

Parameters

[in] solver_options Solver options.

◆ solve()

void robotoc::OCPSolver::solve	(	const double	t,
		const Eigen::VectorXd &	q,
		const Eigen::VectorXd &	v,
		const bool	init_solver = `true`
	)

Solves the optimal control problem. Internally calls updateSolutio() and discretize().

Parameters

[in]	t	Initial time of the horizon.
[in]	q	Initial configuration. Size must be Robot::dimq().
[in]	v	Initial velocity. Size must be Robot::dimv().
[in]	init_solver	If true, initializes the solver, that is, calls discretize(), initConstraints(), and clears the line search filter. Default is true.

Remarks: The linear and angular velocities of the floating base are assumed to be expressed in the body local coordinate.

Friends And Related Function Documentation

◆ operator<<

std::ostream & operator<<	(	std::ostream &	os,
		const OCPSolver &	ocp_solver
	)

friend

The documentation for this class was generated from the following file:

/github/workspace/include/robotoc/solver/ocp_solver.hpp

Public Member Functions

Friends

Detailed Description

Constructor & Destructor Documentation

◆ OCPSolver() [1/4]

◆ OCPSolver() [2/4]

◆ ~OCPSolver()

◆ OCPSolver() [3/4]

◆ OCPSolver() [4/4]

Member Function Documentation

◆ discretize()

◆ disp()

◆ getLQRPolicy()

◆ getRiccatiFactorization()

◆ getSolution() [1/3]

◆ getSolution() [2/3]

◆ getSolution() [3/3]

◆ getSolverStatistics()

◆ getTimeDiscretization()

◆ initConstraints()

◆ KKTError() [1/2]

◆ KKTError() [2/2]

◆ operator=() [1/2]

◆ operator=() [2/2]

◆ setRobotProperties()

◆ setSolution() [1/2]

◆ setSolution() [2/2]

◆ setSolverOptions()

◆ solve()

Friends And Related Function Documentation

◆ operator<<