Play chase with your YetiBorg v2
Robots can make really fun pets and YetiBorg v2 is no exception. The best thing about pets is being able to play games and teach tricks, so we thought we would teach YetiBorg v2 how to follow us like a dog :)
With this example you can see how we can get your robot to follow you around using only the camera.
All we need for this script to work is:
- A YetiBorg v2
- A Raspberry Pi camera so the robot can see
- Someone to chase :P
How does it work?
The script works by taking images from the camera and looking to see if there is any motion. It does this by taking a pair of images and looking at how they are different. For example:
The left image is first, followed by the middle image. The right image is what we get when we take the difference between the two images. We see black where there are no changes.
When we see enough change we get YetiBorg v2 to move forward. In order to decide which way we are turning we take the same difference image and slice it up into sections.
On the left we have the difference image. In the middle we have sliced up the image into our sections. On the right we have chosen the section with the most differences and decided to aim for that. The further it is from the center the faster we will turn left or right.
This all sounds good, but what happens once we are already moving?
As you can see we get lots of movement and YetiBorg v2 will get confused :(
We have a way to fix this though. If you look at the first image you can see movement only in some sections. When YetiBorg v2 is driving we see movement in all sections. What we do then is take the average amount of change in all sections and use that as our baseline for no movement. Now he will not drive from the above images.
But does this still work when both moving and following someone?
It does, the robot can still see a much larger difference where someone is moving then it sees for the movement of the background.
Tweaking the behaviour
Like any good dog, YetiBorg v2 will often become distracted by other things and forget whom to follow. We can improve the behaviour by tweaking some of the settings used for detecting movement in the "Auto drive settings" section.
This is the number of slices the image is split into for detection. Either too high or too low can cause detection problems. We have found a value somewhere between 40 and 120 works best.
Changing this value will adjust how much motion is needed before YetiBorg v2 starts chasing. If the robot does not follow you at all try turning this down. If the robot chases inanimate objects a lot (chairs, tables, et cetera) you may need to turn it up.
This controls how fast YetiBorg v2 will turn to face movement. Smaller values may not react fast enough and larger values will often case the robot to wag instead of follow :) Turn this up if you cannot get the robot to keep up with your movement.
If the robot seems to be running away from movement he may have is camera flipped upside-down. Swap this setting between
Falseto change which way up the images are to fix the problem.
There are also some other things which cannot be fixed by tweaking some numbers:
- The camera has a narrow field of view, making it easy to hide from YetiBorg v2 by standing too far left or right.
- Like a T-Rex YetiBorg v2 only sees movement. If you stand perfectly still you will be invisible :)
- Walking pace is best, very fast movements can also be missed.
- May not play well with other pets. You have been warned :D
- Works best in a well lit area. The camera cannot see much change with someone dressed in black in a dark room :(
- Very tall objects may still attract attention, such as lamp posts...
Get the example
The example is part of the standard set of YetiBorg v2 examples installed during the getting started instructions:
bash <(curl https://www.piborg.org/installer/install-yetiborg-v2.txt)
Go to the YetiBorg v2 code directory:
cd ~/yetiborgv2 and run the script using:
Run at startup
Open /etc/rc.local to make an addition using:
sudo nano /etc/rc.local Then add this line just above the
exit 0 line:
.py & Finally press CTRL+O, ENTER to save the file followed by CTRL+X to exit nano. Next time you power up the Raspberry Pi it should start the script for you :)
Full code listing - yeti2FollowMe.py
#!/usr/bin/env python # coding: Latin-1 # Load library functions we want import time import os import sys import ZeroBorg import io import threading import picamera import picamera.array import cv2 import numpy # Re-direct our output to standard error, we need to ignore standard out to hide some nasty print statements from pygame sys.stdout = sys.stderr print 'Libraries loaded' # Global values global running global ZB global camera global processor global motionDetected running = True motionDetected = False # Setup the ZeroBorg ZB = ZeroBorg.ZeroBorg() #ZB.i2cAddress = 0x44 # Uncomment and change the value if you have changed the board address ZB.Init() if not ZB.foundChip: boards = ZeroBorg.ScanForZeroBorg() if len(boards) == 0: print 'No ZeroBorg found, check you are attached :)' else: print 'No ZeroBorg at address %02X, but we did find boards:' % (ZB.i2cAddress) for board in boards: print ' %02X (%d)' % (board, board) print 'If you need to change the I²C address change the setup line so it is correct, e.g.' print 'ZB.i2cAddress = 0x%02X' % (boards) sys.exit() #ZB.SetEpoIgnore(True) # Uncomment to disable EPO latch, needed if you do not have a switch / jumper # Ensure the communications failsafe has been enabled! failsafe = False for i in range(5): ZB.SetCommsFailsafe(True) failsafe = ZB.GetCommsFailsafe() if failsafe: break if not failsafe: print 'Board %02X failed to report in failsafe mode!' % (ZB.i2cAddress) sys.exit() ZB.ResetEpo() # Power settings voltageIn = 8.4 # Total battery voltage to the ZeroBorg (change to 9V if using a non-rechargeable battery) voltageOut = 6.0 # Maximum motor voltage # Camera settings imageWidth = 320 # Camera image width imageHeight = 240 # Camera image height frameRate = 10 # Camera image capture frame rate # Auto drive settings autoZoneCount = 80 # Number of detection zones, higher is more accurate autoMinimumMovement = 20 # Minimum movement detection before driving steeringGain = 4.0 # Use to increase or decrease the amount of steering used flippedImage = True # True if the camera needs to be rotated showDebug = True # True to display detection values # Setup the power limits if voltageOut > voltageIn: maxPower = 1.0 else: maxPower = voltageOut / float(voltageIn) # Calculate the nearest zoning which fits zones = range(0, imageWidth, imageWidth / autoZoneCount) zoneWidth = zones zoneCount = len(zones) # Image stream processing thread class StreamProcessor(threading.Thread): def __init__(self): super(StreamProcessor, self).__init__() self.stream = picamera.array.PiRGBArray(camera) self.event = threading.Event() self.lastImage = None self.terminated = False self.reportTick = 0 self.start() self.begin = 0 def run(self): # This method runs in a separate thread while not self.terminated: # Wait for an image to be written to the stream if self.event.wait(1): try: # Read the image and do some processing on it self.stream.seek(0) self.ProcessImage(self.stream.array) finally: # Reset the stream and event self.stream.seek(0) self.stream.truncate() self.event.clear() # Image processing function def ProcessImage(self, image): # Flip the image if needed if flippedImage: image = cv2.flip(image, -1) # If this is the first image store and move on if self.lastImage is None: self.lastImage = image.copy() return # Work out the difference from the last image imageDiff = cv2.absdiff(self.lastImage, image) # Build up the zone change levels zoneDetections =  for zone in zones: # Grab the zone from the differences zoneDiff = imageDiff[:, zone : zone + zoneWidth, :] # Get an average for the zone zoneChange = zoneDiff.mean() zoneDetections.append(zoneChange) # Set drives or report motion status self.SetSpeedFromDetection(zoneDetections) # Save the previous image self.lastImage = image.copy() # Set the motor speed from the motion detection def SetSpeedFromDetection(self, zoneDetections): global ZB global motionDetected # Find the largest and average detections largestZone = 0 largestDetection = 0 averageDetection = 0 for i in range(zoneCount): if zoneDetections[i] > largestDetection: largestZone = i largestDetection = zoneDetections[i] averageDetection += zoneDetections[i] averageDetection /= float(zoneCount) # Remove the baseline motion from the largest zone detection = largestDetection - averageDetection # Determine if the motion is strong enough to count as a detection if detection > autoMinimumMovement: # Motion detected motionDetected = True if showDebug: if self.reportTick < 2: print 'MOVEMENT %05.2f [%05.2f %05.2f]' % (detection, largestDetection, averageDetection) print ' Zone %d of %d' % (largestZone + 1, zoneCount) self.reportTick = frameRate else: self.reportTick -= 1 # Calculate speeds based on zone steering = ((2.0 * largestZone) / float(zoneCount - 1)) - 1.0 steering *= steeringGain if steering < 0.0: # Steer to the left driveLeft = 1.0 + steering driveRight = 1.0 if driveLeft <= 0.05: driveLeft = 0.05 else: # Steer to the right driveLeft = 1.0 driveRight = 1.0 - steering if driveRight <= 0.05: driveRight = 0.05 else: # No motion detected motionDetected = False if showDebug: if self.reportTick < 2: print '-------- %05.2f [%05.2f %05.2f]' % (detection, largestDetection, averageDetection) self.reportTick = frameRate else: self.reportTick -= 1 # Stop moving driveLeft = 0.0 driveRight = 0.0 # Set the motors ZB.SetMotor1(-driveRight * maxPower) # Rear right ZB.SetMotor2(-driveRight * maxPower) # Front right ZB.SetMotor3(-driveLeft * maxPower) # Front left ZB.SetMotor4(-driveLeft * maxPower) # Rear left # Image capture thread class ImageCapture(threading.Thread): def __init__(self): super(ImageCapture, self).__init__() self.start() def run(self): global camera global processor print 'Start the stream using the video port' camera.capture_sequence(self.TriggerStream(), format='bgr', use_video_port=True) print 'Terminating camera processing...' processor.terminated = True processor.join() print 'Processing terminated.' # Stream delegation loop def TriggerStream(self): global running while running: if processor.event.is_set(): time.sleep(0.01) else: yield processor.stream processor.event.set() # Startup sequence print 'Setup camera' camera = picamera.PiCamera() camera.resolution = (imageWidth, imageHeight) camera.framerate = frameRate imageCentreX = imageWidth / 2.0 imageCentreY = imageHeight / 2.0 print 'Setup the stream processing thread' processor = StreamProcessor() print 'Wait ...' time.sleep(2) captureThread = ImageCapture() try: print 'Press CTRL+C to quit' ZB.MotorsOff() # Loop indefinitely while running: # # Change the LED to show if we have detected motion # We do this regularly to keep the communications failsafe test happy ZB.SetLed(motionDetected) # Wait for the interval period time.sleep(0.1) # Disable all drives ZB.MotorsOff() except KeyboardInterrupt: # CTRL+C exit, disable all drives print ' User shutdown' ZB.MotorsOff() except: # Unexpected error, shut down! e = sys.exc_info() print print e print ' Unexpected error, shutting down!' ZB.MotorsOff() # Tell each thread to stop, and wait for them to end running = False captureThread.join() processor.terminated = True processor.join() del camera ZB.SetLed(False) print 'Program terminated.'
Leave a Comment
Leave a Reply