RTIS Dev

This project is a Python Module to be used on an embedded platform connected to a RTIS sensor. It can be used for quick prototyping and developing Python scripts related to the sensor data from RTIS sensors. It includes functions to measure and process sonar data.

Installation

Dependencies

Version numbers are those used during this project, higher versions likely supported too unless otherwise specified.

• Python 3.6 or higher with modules:
  • Numpy
  • Scipy
  • multiprocessing-logging
  • uuid
  • RTIS Serial on Linux systems (see here)
  • PySerial on Windows systems
  • RTIS CUDA 2.6.1 or higher if any processing is desired (see here)

Install from PyPI

You can install this module from the PyPi repository like any other:

pip install rtisdev

Install from source

A. Automatic installation using RTIS Update

To build and install RTIS Dev and keep it automatically up to date, please use the RTIS Update script.

B. Manual Installation

Install all dependencies listed above. We suggest to always use pip.

pip install MODULE_NAME

To install the Python Module so that you can use it in any Python script on Windows:

py -m pip install .

And on Linux:

pip install .

Usage

All methods and classes are fully documented on the wiki page. which has more information and examples.

1) Import

If rtisdev is installed you can now import it in any Python script on the system as any other module:

import rtisdev

Do note that due to RTIS Dev using RTIS CUDA, you will need to add the folder where you installed RTIS CUDA to your library path for your script if you are using an IDEA that uses custom Python environments. For example, using PyCharm IDE can have the issue of not being able to load RTIS CUDA library by not finding library path: You will need to add LD_LIBRARY_PATH with value /usr/lib/rtiscuda (or your custom location) manually to your run configurations.

2) General Usage

There are several classes and methods available in RTIS Dev. Some general knowledge on RTIS sonar sensors is advised. Here is a small example that goes over most basic steps:

import rtisdev
import matplotlib.pyplot as plt
import numpy as np

# Open a connection to the RTIS Device over the default serial port.
success_connect = rtisdev.open_connection(allowDebugMode=True)

# Set the recording settings with 163840 samples and a call sweep between 25 and 50 KHz.
config_uuid = rtisdev.set_recording_settings(microphoneSamples=163840, callMinimumFrequency=25000, callMaximumFrequency=50000)

# Enable all processing steps and preload them with RTIS CUDA. This will produce a 2D energyscape with 91 directions
# with a maximum distance of 6m.
success_settings_processing = rtisdev.set_processing_settings(directions=91, maxRange=6, configName=config_uuid)

# Get the used settings as a RTISSettings object.
settings = rtisdev.get_current_settings(configName=config_uuid)

# Get an ACTIVE measurement (protect your ears!) in raw data.
measurement = rtisdev.get_raw_measurement(True, configName=config_uuid)

# Store the raw data of that measurement as a binary file. This can be opened in another application for further work.
# For an example in MATLAB, see below.
raw_data_sonar = measurement.rawData.tobytes()
file_handle_data = open("test_measurement_" + str(measurement.index) + ".bin", "wb")
file_handle_data.write(raw_data_sonar)
file_handle_data.close()

# Process that raw measurement to an energyscape using the configuration chosen earlier.
processed_measurement = rtisdev.process_measurement(measurement, configName=config_uuid)

# Get a new ACTIVE measurement (protect your ears!) in both raw and processed data formats directly.
new_processed_measurement = rtisdev.get_processed_measurement(True, configName=config_uuid)

# Plot the 2D energyscape of this processed measurement.
plt.imshow(np.transpose(new_processed_measurement.processedData), cmap="hot", interpolation='nearest')
plt.xlabel("Directions (degrees)")
plt.ylabel("Range (meters)")
indexes_x = np.arange(0, new_processed_measurement.processedData.shape[0], 20)
labels_x = np.round(np.rad2deg(settings.directions[indexes_x, 0])).astype(int)
indexes_y = np.arange(0, new_processed_measurement.processedData.shape[1], 100)
labels_y = settings.ranges[indexes_y]
fmt_x = lambda x: "{:.0f}°".format(x)
fmt_y = lambda x: "{:.2f}m".format(x)
plt.xticks(indexes_x, [fmt_x(i) for i in labels_x])
plt.yticks(indexes_y, [fmt_y(i) for i in labels_y])
plt.title("RTIS Dev - 2D Energyscape Example")
ax = plt.gca()
ax.invert_yaxis()
ax.set_aspect("auto")
plt.show()

# Get a new ACTIVE measurement (protect your ears!) in both raw and microphone signal format directly.
signal_measurement = rtisdev.get_signal_measurement(True, configName=config_uuid)

# Plot the microphone signals of this measurement.
fig, axs = plt.subplots(8, 4, figsize=(10,16), constrained_layout = True)
for microphone_index_i in range(0, 8):
   for microphone_index_j in range(0, 4):
       axs[microphone_index_i, microphone_index_j].set_title(str(microphone_index_j+(microphone_index_i*4)+1))
       axs[microphone_index_i, microphone_index_j].plot(signal_measurement.processedData[microphone_index_j+(microphone_index_i*4),:])
       if microphone_index_j != 0:
           plt.setp(axs[microphone_index_i, microphone_index_j], yticklabels=[])
       if microphone_index_i != 7:
           plt.setp(axs[microphone_index_i, microphone_index_j], xticklabels=[])
       if microphone_index_i == 7:
           axs[microphone_index_i, microphone_index_j].set_xlabel("Time (Samples)")
       if microphone_index_j == 0:
           axs[microphone_index_i, microphone_index_j].set_ylabel("Amplitude")
fig.suptitle("RTIS Dev - Microphone Signals")
plt.show()

In the example above, a binary file is saved of the recorded measurement. This can be opened in another application for further work. In the example below, MATLAB is used to process the raw data to microphone signals and plot these.

% Read binary data
file_name = 'test_measurement_0.bin';
f_id = fopen(file_name, 'rb');
rawData = fread(f_id, 'uint32');
fclose(f_id);
data =  de2bi(rawData, 32);

% Generate settings for converting to microphone signals. These have to match the recording settings!
fs_pdm = 450e4; % PDM samplefrequency
fs_mic = 450e3; % Microphone samplefrequency
r_min = 0.5; % Range minimum to show
r_max = 5; % Range maximum to show

% PDM demodulation.
[b_pdm, a_pdm] = butter(6, 100e3 / ( fs_pdm / 2));
[b_bpf, a_bpf] = butter(6, [20e3 80e3] / (fs_mic / 2));
data_filtered = (filter(b_pdm, a_pdm, data));
data_filtered_dec = data_filtered(1:10:end, :);
data_mics = filter(b_bpf, a_bpf, data_filtered_dec);

% Subselect the part between the minimum and maximum range wanted.
spl_start = round(r_min * 2 / 343 * fs_mic); 
spl_stop = round(r_max * 2 / 343 * fs_mic);
data_mics = data_mics(spl_start : spl_stop, :);
data_mics = data_mics - mean(data_mics);

%% Plot the recording microphone signals
figure()
for plotcounter = 1 : 16
    subplot(4, 4, plotcounter);
    sig_out1 = data_mics(:, plotcounter) - mean(data_mics(:,plotcounter));
    plot(sig_out1 + 1);
    hold on;
    sig_out2 = data_mics(:, plotcounter + 16) - mean(data_mics(:, plotcounter + 16));
    plot(sig_out2  - 1);
    ylim([-1.5 1.5])
    hold off;
    plottitle = sprintf('Microphone %d and %d', (plotcounter * 2) - 1, plotcounter * 2);
    title(plottitle);
end