Signals for control tasks

This is a small package to provide Signal objects that can be evaluated at time t. The package only relies on numpy and implements arithmetic operations between Signal types, int and float primitives.

Example Usage

Add two signals together and evaluate at a single time-step

from signals import Step, Sinusoid

step = Step(t_start=2.0, t_end=20.0)
sine = Sinusoid(t_start=0.0, t_end=10.0, freq=2.0)
reference = 5 * step + sine

# Reference at t = 5.0:
r_t = reference(t=5.0)

Every Signal object has t_start and a t_end arguments. In most cases the default t_start and t_end are set to 0 and +inf respectively. For any t not in [t_start, t_end] the Signal object returns 0.

Make a sequence of step signals and evaluate them on an array time-signals:

import numpy as np
from signals import StepSequence

amplitudes = [2, 1, 0, -1, 2, 0]
times = [1, 2, 3, 4, 5, 7]
step_sequence = StepSequence(times=times, amplitudes=amplitudes)

# Time array:
t = np.arange(0, 10.0, 0.1)
s = step_sequence.eval_on(t=t)

Installation

1. Manually using pip

pip install git+https://github.com/aerospace-rl/signals.git

2. In local virtual environment in edit mode

git clone https://github.com/aerospace-rl/signals.git`
cd signals
pip install -e .

3. Automatically using conda and environment.yml:

dependencies:
  - pip
  - pip:
    - git+https://github.com/aerospace-rl/signals.git

Defining custom signals

In order to make your own signal, make a class that derives from Signal and define the _signal(self, t: float) -> float method.

from signals import Signal

class NegativeRamp(Signal):
    def _signal(self, t: float) -> float:
        return -t

To-do / Feature list

• unit testing
• CI with unit tests and black formatting