REBOTNIX SDK

Welcome to the REBOTNIX RB-SDK | Software Development Kit

You can use our SDK to build your own custom AI pipelines, communicate with our hardware modules like GPS, Lidar, Modbus, MPU or ToF sensors. The SDK includes different camera types, frame grabbing, extraction of GPS-Coordinates and Object-Detection-Inference on images.

The SDK is simple to use and ready-to-go on our REBOTNIX GUSTAV edge devices which all run on NVIDIA GPUs.

OPENCV – CUDA FOR JETPACK´S: 4.6.1 – 5.x or Higher.

Currently we offer bindings in C,Python 3.7 or higher.

RB_CAM_BASLER

The REBOTNIX RB_CAMERA_BASLER AND RB_CAMERA_AV is a tool to grab frames from a Basler and Allied VISION cameras. Please note that this only works on preflashed devices from us.

Requirements for python 3.7:
numpy
opencv-python
pycrypto
pypylon

Get camera list

The camera class can be used to list all available cameras that are connected to the machine.

We do support USB- as well as IP-cameras, e.g GigE.

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

# import rb_camera
import rb_camera_basler as rb_c

#get list of available cameras
available_cameras = rb_c.listCameras()

print(available_cameras)

The above command returns a list structured like this:

[
  {
    "class": "BaslerUsb",
    "modelname": "a2A1920-160ucBAS",
    "serial": "40087915",
    "ip": None
  }
]

Initialization

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

# import rb_camera
import rb_camera_basler as rb_c

# initialize rb_camera module
rb_cam = rb_c.rb_camera(serial="40087915")

#stop frame grabbing.
rb_cam.stop()

Parameters

Parameter	Default	Description
serial		the camera serial number.
overwrite	True	If set to true, the configurations will be overwritten.
width	1920	Pixel-width of camera.
height	1080	Pixel-height of camera.
exposureTime	1000	Exposure time in milliseconds.
GainAuto	"Once"	Auto-Gain mode for Basler camera.
pixelFormat	"BayerRG8"	The pixel color format of the grabbed frame.
fps	25	Frames per second.
exposureMin	27	Lower limit for exposure time.
exposureMax	35000	Upper limit for exposure time.
lightarget	0.35	Target brightness value you want to keep.
gpioreset	False	Do a gpio reset.
usbreset	False	Do an USB reset.

Grab new frame

Once you have the camera serial you can initialize the camera to be ready for frame grabbing.

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#REBOTNIX Camera SDK
import libs.rb_camera_basler as rb_c

#get a vailable cameras
available_cameras = rb_c.listCameras()
print(available_cameras)

#set serial of basler camera
rb_cam = rb_c.rb_camera(available_cameras[0]["serial"], usbreset=False)

while True:
  # read current frame
  flag, frame = rb_cam.getFlagAndFrame()
  if flag:
    print(frame.shape)

rb_cam.stop()

Outputs

Output	Type	Description
flag	Boolean	Flag is true if a frame was grabbed succesfully, false if not.
frame	numpy.ndarray	The grabbed frame as a numpy array.

Download

Download rb_camera_basler

RB_CAM_AV

The REBOTNIX RB_CAMERA_ALLIEDVISION is a tool to grab frames from a Alliedvision camera.

Requirements for python 3.7:
vimba
numpy

Get camera list

The camera class can be used to list all available cameras that are connected to the machine.

We do support USB- as well as IP-cameras, e.g GigE.

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

# import rb_camera
import rb_camera_alliedvision as rb_c

#get list of available cameras
available_cameras = rb_c.listCameras()

print(available_cameras)

The above command returns a list structured like this:

[
  {
    "class": "AlliedvisionUsb",
    "modelname": "Alvium 1500 C-050",
    "serial": "0266J",
    "ip": None
  }
]

Initialization

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

# import rb_camera
import rb_camera_alliedvision as rb_c

# initialize rb_camera module
rb_cam = rb_c.rb_camera(serial="0266J")

#stop frame grabbing.
rb_cam.stop()

Parameters

Parameter	Default	Description
serial		The camera serial number.
exposureMin	27	Lower limit for exposure time.
exposureMax	35000	Upper limit for exposure time.
gpioreset	False	Do a gpio reset.
usbreset	False	Do an USB reset.
xml_config		Path to an xml-config file

Grab new Frame

Once you have the camera serial you can initialize the camera to be ready for frame grabbing.

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#os.environ["GENICAM_GENTL64_PATH"] = ":/opt/pylon/lib/gentlproducer/gtl:/opt/ssd/install_vimba/VimbaGigETL/CTI/arm_64bit:/opt/ssd/install_vimba/VimbaUSBTL/CTI/arm_64bit"

#REBOTNIX Camera SDK
import libs.rb_camera_alliedvision as rb_c

#get a vailable cameras
available_cameras = rb_c.listCameras()
print(available_cameras)
rb_cam = rb_c.rb_camera(available_cameras[0]["serial"], usbreset=False)

while True:
  # read current frame
  flag, frame = rb_cam.getFlagAndFrame()
  if flag:
    print(frame.shape)

rb_cam.stop()

Outputs

Output	Type	Description
flag	Boolean	Flag is true if a frame was grabbed succesfully, false if not.
frame	numpy.ndarray	The grabbed frame as a numpy array.

Download

Download rb_camera_alliedvision

RB_GPS

The REBOTNIX RB_GPS is a tool to extract raw data from a UBLOX-GPS or other compatible module.

Initialization

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#import rb_gps
from libs.rb_gps import rb_gps 

# initialize and start rb_gps module
mygps = rb_gps()

#check if valid gps
print(mygps.flag)

# stop rb_gps module
mygps.stop()

The above commands start and stop the rb_gps module. If you want to set the GPS Module starting with non root, you have to set some flags.

a) sudo chmod a+rw /dev/ttyACM0

b) sudo usermod -a -G dialout rebotnix

c) reboot.

Now the GPS module can be accessed with another user as root.

Get GPS meta data

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

import time

#REBOTNIX GPS SDK
from libs.rb_gps import rb_gps
#initialize gps
mygps = rb_gps()

while True:
    if mygps.flag:
        lon, lat, gps_acc, sat, speed, headAcc, headMot = mygps.getgps()
        print(lon, lat, gps_acc, sat, speed, headAcc, headMot)
        time.sleep(0.1)

mygps.stop()

Outputs | Output | Type | Description | | ------------ | ------------- | ------------ | |lat | float | The current latitude as a float. | | lon| float | The current longitude as a float. | |gps_acc | float | The accuracy of the signal. | |sat | float | The number of satelites.| |speed | float | Speed information.| |headAcc | float | Accuracy of heading. | |headMot | float | Mot of heading. |

Download

Download rb_gps

RB_KINEMATICS

The REBOTNIX RB_KINEMATICS is a tool to extract raw data from a RTK-GPS or other compatible module.

Initialization

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#import rb_kinematics
from libs.rb_kinematics import rb_kinematics

# initialize and start rb_kinematics module
mykinematics = rb_kinematics(port="/dev/ttyACM1")

#check if valid information
print(mykinematics.flag)

# stop rb_gps module
mykinematics.stop()

Get Kinematics meta data

import sys
import os

#check for ROOT_PATH variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

import time

#import rb_kinematics
from libs.rb_kinematics import rb_kinematics

# initialize and start rb_kinematics module
mykinematics = rb_kinematics(port="/dev/ttyACM1")

while True:
    if mykinematics.flag:
        lon, lat, gps_acc, sat, speed, headAcc, headMot = mykinematics.getgps()
        print(lon, lat, gps_acc, sat, speed, headAcc, headMot)
        time.sleep(0.1)

mykinematics.stop()

Outputs

Output	Type	Description
lat	float	The current latitude as a float.
lon	float	The current longitude as a float.
gps_acc	float	The accuracy of the signal.
sat	float	The number of satelites.
speed	float	Speed information.

Download

Download rb_kinematics

RB_VISIONTOOLS

The REBOTNIX RB_VISIONTOOLS is a tool to detect objects using a custom trained Object Detection Model.

Since it natevily runs on TRT it offers a very high frames-per-second (FPS) rate. It therefore enables you to build real-time applications. This module runs only on a NVIDIA embedded GPU.

Load Model

import sys
import os

#check for root_path variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#import rb_visiontools
import libs.rb_visiontools as rb_visiontools

model_path = "./rebotnix_model.rvt"
# load rebotnix model
network, class_names, class_colors, width, height = rb_visiontools.load_model_rb(model_path)

|Parameter |Type | Description| | ------------ | ------------- | ------------ | |model_path|String|Path to model container (rvt file).|

The above command returns the rb_model as well as the class-names, the class-colors and the rb_model width and height.

Inference

import sys
import os

#check for root_path variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#import rb_visiontools
import libs.rb_visiontools as rb_visiontools
import cv2

model_path = "./models/rebotnix_model.rvt"
# load rebotnix model
network, class_names, class_colors, width, height = rb_visiontools.load_model_rb(model_path)
# set detection threshold
threshold = 0.75

frame_path = "demo.jpg"
frame = cv2.imread(frame_path)

# run detection on input frame
detections, scale_x, scale_y = rb_visiontools.do_detection_rb(network, class_names, frame, width, height, threshold)#

print(detections)

#Output: 
#[('206', '99.1', (289.4804382324219, 107.32664489746094, 107.9014663696289, 171.51626586914062))]

Inputs

Parameter	Type	Description
network	rb_network	Loaded model.
class_names	list	List of detection classes.
frame	numpy.ndarray	Input frame to run inference on.
width	int	Width of rb_model.
height	int	Height of rb_model.
threshold	float	Threshold for inference. Between 0 and 1.

Download

Download rb_visiontools

RB_MODBUS

The REBOTNIX RB_MODBUS is a tool to connect to a modbus server. It enables you to read and process machine data in real time.

Reading Data

import sys
import os

#check for root_path variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

#import rb_modbus
import libs.rb_modbusClass as rb_modbus

#connect to modbus with host ip and port 
host = "192.168.1.1"
port = 1502
modbus_connection = rb_modbus.rb_modbusClass(host=host,port=port)

#start connection to modbus
modbus_connection.start()

while True:

  if modbus_connection.error == True:
    break

  #check if new signal is coming in
  if modbus_connection.rb_lock:

    #read signal
    message = modbus_connection.rb_words

    print(message)

#stop connection to modbus
modbus_connection.stop()

Parameters

Parameter	Type	Description
host	String	IP address of your modbus server.
port	int	Port number.

Download

Download rb_modbus

RB_COMPRESS

REBOTNIX RB_COMPRESS is a tool for compressing relevant information for transferring images to the smallest possible file size. By reducing the file size, you save data volume, time and cost in applications such as smart city applications, robotics or any other IoT without losing information.

Requirements for python 3.7:
opencv-python

Compressing an image

import os
import sys

#check for root_path variable
try:
  sys.path.append(os.environ['ROOT_PATH'])
except:
  print("Please set the variable ROOT_PATH.")
  sys.exit(0)

import cv2
import libs.rb_compressClass as rb_comp

licpath = "./rb_compress.lic"

# input image
img = cv2.imread("./demo.jpg")

# if you want, you can do a resize as well
# aspect ratio

#ar = 1920 / img.shape[1]
#dim = (1920, int(img.shape[0] *ar))
#img = cv2.resize(img,dim, interpolation = cv2.INTER_AREA)

bitrate="45000k"
speed='high'
out_name="./demo.avif"

#init compress module
compressor = rb_comp.rb_compress(licpath=licpath)

#run compression
compressor.compress(img, bitrate, speed, out_name)

Parameters

Parameter	Type	Description
licpath	String	Path of your license file.
img	Image	Image object.
quality	String	quality parameter, [“low”,”mid”,”high”].
output_name	String	Path of your desired output file.

Download

Download rb_compress

GPIO RESET

Here is shell code to reset the gpio.

echo 442 > /sys/class/gpio/export
sleep 1

echo 444 > /sys/class/gpio/export
sleep 1

echo out > /sys/class/gpio/gpio442/direction
sleep 1

echo out > /sys/class/gpio/gpio444/direction

# turn off
sleep 3
echo 1 > /sys/class/gpio/gpio442/value
echo 1 > /sys/class/gpio/gpio444/value

sleep 3

# turn on
echo 0 > /sys/class/gpio/gpio442/value
echo 0 > /sys/class/gpio/gpio444/value