# forwardDynamics

Joint accelerations given joint torques and states

## Syntax

``jointAccel = forwardDynamics(robot)``
``jointAccel = forwardDynamics(robot,configuration)``
``jointAccel = forwardDynamics(robot,configuration,jointVel)``
``jointAccel = forwardDynamics(robot,configuration,jointVel,jointTorq)``
``jointAccel = forwardDynamics(robot,configuration,jointVel,jointTorq,fext)``

## Description

````jointAccel = forwardDynamics(robot)` computes joint accelerations due to gravity at the robot home configuration, with zero joint velocities and no external forces.```
````jointAccel = forwardDynamics(robot,configuration)` also specifies the joint positions of the robot configuration.To specify the home configuration, zero joint velocities, or zero torques, use `[]` for that input argument.```
````jointAccel = forwardDynamics(robot,configuration,jointVel)` also specifies the joint velocities of the robot.```
````jointAccel = forwardDynamics(robot,configuration,jointVel,jointTorq)` also specifies the joint torques applied to the robot. ```

example

````jointAccel = forwardDynamics(robot,configuration,jointVel,jointTorq,fext)` also specifies an external force matrix that contains forces applied to each joint.```

## Examples

collapse all

Calculate the resultant joint accelerations for a given robot configuration with applied external forces and forces due to gravity. A wrench is applied to a specific body with the gravity being specified for the whole robot.

Load a predefined KUKA LBR robot model, which is specified as a `RigidBodyTree` object.

`load exampleRobots.mat lbr`

Set the data format to `'row'`. For all dynamics calculations, the data format must be either `'row'` or `'column'`.

`lbr.DataFormat = 'row';`

Set the gravity. By default, gravity is assumed to be zero.

`lbr.Gravity = [0 0 -9.81];`

Get the home configuration for the `lbr` robot.

`q = homeConfiguration(lbr);`

Specify the wrench vector that represents the external forces experienced by the robot. Use the `externalForce` function to generate the external force matrix. Specify the robot model, the end effector that experiences the wrench, the wrench vector, and the current robot configuration. `wrench` is given relative to the `'tool0'` body frame, which requires you to specify the robot configuration, `q`.

```wrench = [0 0 0.5 0 0 0.3]; fext = externalForce(lbr,'tool0',wrench,q);```

Compute the resultant joint accelerations due to gravity, with the external force applied to the end-effector `'tool0'` when `lbr` is at its home configuration. The joint velocities and joint torques are assumed to be zero (input as an empty vector `[]`).

`qddot = forwardDynamics(lbr,q,[],[],fext);`

## Input Arguments

collapse all

Robot model, specified as a `rigidBodyTree` object. To use the `forwardDynamics` function, set the `DataFormat` property to either `'row'` or `'column'`.

Robot configuration, specified as a vector with positions for all nonfixed joints in the robot model. You can generate a configuration using `homeConfiguration(robot)`, `randomConfiguration(robot)`, or by specifying your own joint positions. To use the vector form of `configuration`, set the `DataFormat` property for the `robot` to either `'row'` or `'column'`.

Joint velocities, specified as a vector. The number of joint velocities is equal to the degrees of freedom of the robot. To use the vector form of `jointVel`, set the `DataFormat` property for the `robot` to either `'row'` or `'column'`.

Joint torques, specified as a vector. Each element corresponds to a torque applied to a specific joint. To use the vector form of `jointTorq`, set the `DataFormat` property for the `robot` to either `'row'` or `'column'`.

External force matrix, specified as either an n-by-6 or 6-by-n matrix, where n is the number of bodies of the robot. The shape depends on the `DataFormat` property of `robot`. The `'row'` data format uses an n-by-6 matrix. The `'column'` data format uses a 6-by-n .

The matrix lists only values other than zero at the locations relevant to the body specified. You can add force matrices together to specify multiple forces on multiple bodies.

To create the matrix for a specified force or torque, see `externalForce`.

## Output Arguments

collapse all

Joint accelerations, returned as a vector. The dimension of the joint accelerations vector is equal to the degrees of freedom of the robot. Each element corresponds to a specific joint on the `robot`.

collapse all

### Dynamics Properties

When working with robot dynamics, specify the information for individual bodies of your manipulator robot using these properties of the `rigidBody` objects:

• `Mass` — Mass of the rigid body in kilograms.

• `CenterOfMass` — Center of mass position of the rigid body, specified as a vector of the form `[x y z]`. The vector describes the location of the center of mass of the rigid body, relative to the body frame, in meters. The `centerOfMass` object function uses these rigid body property values when computing the center of mass of a robot.

• `Inertia` — Inertia of the rigid body, specified as a vector of the form `[Ixx Iyy Izz Iyz Ixz Ixy]`. The vector is relative to the body frame in kilogram square meters. The inertia tensor is a positive definite matrix of the form: The first three elements of the `Inertia` vector are the moment of inertia, which are the diagonal elements of the inertia tensor. The last three elements are the product of inertia, which are the off-diagonal elements of the inertia tensor.

For information related to the entire manipulator robot model, specify these `rigidBodyTree` object properties:

• `Gravity` — Gravitational acceleration experienced by the robot, specified as an `[x y z]` vector in m/s2. By default, there is no gravitational acceleration.

• `DataFormat` — The input and output data format for the kinematics and dynamics functions, specified as `"struct"`, `"row"`, or `"column"`.

### Dynamics Equations

Manipulator rigid body dynamics are governed by this equation:

`$\frac{d}{dt}\left[\begin{array}{c}q\\ \stackrel{˙}{q}\end{array}\right]=\left[\begin{array}{c}\stackrel{˙}{q}\\ M{\left(q\right)}^{-1}\left(-C\left(q,\stackrel{˙}{q}\right)\stackrel{˙}{q}-G\left(q\right)-J{\left(q\right)}^{T}{F}_{Ext}+\tau \right)\end{array}\right]$`

also written as:

`$M\left(q\right)\stackrel{¨}{q}=-C\left(q,\stackrel{˙}{q}\right)\stackrel{˙}{q}-G\left(q\right)-J{\left(q\right)}^{T}{F}_{Ext}+\tau$`

where:

• $M\left(q\right)$ — is a joint-space mass matrix based on the current robot configuration. Calculate this matrix by using the `massMatrix` object function.

• $C\left(q,\stackrel{˙}{q}\right)$ — is the coriolis terms, which are multiplied by $\stackrel{˙}{q}$ to calculate the velocity product. Calculate the velocity product by using by the `velocityProduct` object function.

• $G\left(q\right)$ — is the gravity torques and forces required for all joints to maintain their positions in the specified gravity `Gravity`. Calculate the gravity torque by using the `gravityTorque` object function.

• $J\left(q\right)$ — is the geometric Jacobian for the specified joint configuration. Calculate the geometric Jacobian by using the `geometricJacobian` object function.

• ${F}_{Ext}$ — is a matrix of the external forces applied to the rigid body. Generate external forces by using the `externalForce` object function.

• $\tau$ — are the joint torques and forces applied directly as a vector to each joint.

• $q,\stackrel{˙}{q},\stackrel{¨}{q}$ — are the joint configuration, joint velocities, and joint accelerations, respectively, as individual vectors. For revolute joints, specify values in radians, rad/s, and rad/s2, respectively. For prismatic joints, specify in meters, m/s, and m/s2.

To compute the dynamics directly, use the `forwardDynamics` object function. The function calculates the joint accelerations for the specified combinations of the above inputs.

To achieve a certain set of motions, use the `inverseDynamics` object function. The function calculates the joint torques required to achieve the specified configuration, velocities, accelerations, and external forces.

 Featherstone, Roy. Rigid Body Dynamics Algorithms. Springer US, 2008. DOI.org (Crossref), doi:10.1007/978-1-4899-7560-7.