Sensor Models

This page describes the sensor models used for simulating uncertain measurements. To enable seamless interchangeability of different sensor models, a common interface including a dataclass for sensor outputs are provided. The implemented concrete models are (briefly) presented at the end of this file.

Currently, we only support full state feedback.

---

Dataclass Objects

Measured outputs are stored as instances of the corresponding OutputInterface dataclass. For full state feedback, the measured state can also be stored as an instanceof the corresponding StateInterface dataclass.

The output objects can be converted to numpy arrays using their convert_to_array() method.

---

Overview and Interface Concept

All sensor models implement the same abstract interface, SensorModelInterface.

Every vehicle model must implement the output function: $$ y= h(x,u,\nu) = h_{\mathrm{nominal}}(x,u) + {\nu} $$ where the nominal output function \(h_{\mathrm{nominal}}({x}, {u})\) computes the nominal measurement from the current state \(x\) and control input \(u\). The (random) sensor noise is denoted by \({\nu} \in \mathbb{R}^{\nu}\).

The nominal output function \(h_{\mathrm{nominal}}({x}, {u})\) must be implemented via the method:

def _output_function_cas(x, u)

For public access, the following method is implemented:

def nominal_output(x, u,)

This function serves as a wrapper for \(h_{\mathrm{nominal}}({x}, {u})\) and accepts states and control inputs that are represented as instances of the respective dataclass.

The noisy output function \(h(x,u,\nu)\) can be evaluated via the method

def measure(x, u, rand_noise)

where the default value of the optional input argument rand_noise is True. If rand_noise=False, the output of measure is the output of the nominal output function \(h_{\mathrm{nominal}}({x}, {u})\).

Constructor

SensorModelInterface(
        noise_model,
        state_output_factory,
        dim,
        state_dimension,
        input_dimension,
    )

with the parameters:

Parameters	Variable type	Description
noise_model	UncertaintyModelInterface	Uncertainty model for generating (random) sensor noise.
state_output_factory	StateInputDisturbanceTrajectoryFactoryInterface	Create output/state from numpy array (output of output function).
dim	int	Dimension of the output vector.
state_dimension	int	Dimension of the state vector.
input_dimension	int	Dimension of the control input vector.

---

Implemented Sensor Models

---

Full State Feedback

Nominal output function: $$ h_{\mathrm{nominal}}(x,u) = x $$

Please note that each vehicle model requires a model-specific output dataclass, which are located alongside the respective vehicle model implementation.