My project is to build a software to help drivers achieve ecodriving thanks to optimal control based advices. Consequently, to feed a linear quadratic regulator, vehicle longitudinal equation of motion should be transformed into a state space form.

Considering a state having :

$$ \begin{align*} & \dot{s} = v\\ & \dot{v} = \mathit{u}_{t} - \mathit{u}_{b} - \frac{1}{m} 0.5 \mathit{\rho}_{air} A \mathit{C}_{d}v^2 - g(sin \theta(s) + \mathit{C}_{rr} cos \theta(s))\\ \\ \end{align*} $$


$\mathit{u}_{t}$ is tractive acceleration

$\mathit{u}_{B}$ is braking acceleration

$m$ is vehicle mass

$\mathit{\rho}_{air}$ is air density

$A$ is the vehicle frontal area

$\mathit{C}_{d}$ is the aerodynamic drag coefficient

$v$ is vehicle velocity

$g$ is gravity

$\theta(s)$ is road grade at position s

$\mathit{C}_{rr}$ is coefficient of rolling

How to linearize nonlinear terms $sin \theta(s)$, $cos \theta(s)$ and $v^2$ ?


2 Answers 2


I guess your project is a school project, therefore my answer will be generic.

Consider the nonlinear state space system $$ \dot{x} = f(x,u), $$ with state $x(t) \in \mathbb{R}^n$ and input $u(t) \in \mathbb{R}^k$. The linearized dynamics around the selected operating point $\left(x_0, u_0\right)$ are then given by $$ \dot{\tilde{x}} = A\tilde{x} + B\tilde{u}, $$ with $\tilde{x}(t) = x(t) - x_0$ and $\tilde{u}(t) = u(t) - u_0$, $A = \frac{\partial f(x_0,u_0)}{\partial x}$ and $B = \frac{\partial f(x_0,u_0)}{\partial u}$.

In your case, I would use the state vector $x = \begin{bmatrix}s \\ v\end{bmatrix}$ and input vector $ u = \begin{bmatrix} u_t \\ u_B \\ \theta \end{bmatrix}$.


In the real world cos(theta)=1 and sin(theta)=theta in radians, assuming you aren't rockhopping in a Jeep.

  • $\begingroup$ This answer needs more detail. $\endgroup$
    – Fred
    Dec 14, 2022 at 11:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.