I'm implementing follow me drone. Condition Yaw working fine, but negative X velocity from NED FRAME giving yaw added and therefore giving circular movement on jetson nano gazebo simulation. But on Intel CPU it's working fine. I check dronekit and pymavlink library version is same.

from future import with_statement from future import division from future import absolute_import from os import sys, path from io import open sys.path.append(path.dirname(path.dirname(path.abspath(file))))

-- Dependencies for video processing

import time import math import argparse import cv2 import numpy as np from imutils.video import FPS

-- Dependencies for commanding the drone

from dronekit import connect, VehicleMode from pymavlink import mavutil

-- Function to control velocity of the drone

def send_local_ned_velocity(x, y, z): msg = vehicle.message_factory.set_position_target_local_ned_encode( 0, 0, 0, mavutil.mavlink.MAV_FRAME_BODY_OFFSET_NED, 0b0000111111000111, 0, 0, 0, x, y, z, 0, 0, 0, 0, 0) vehicle.send_mavlink(msg) vehicle.flush()

def condition_yaw(heading): msg = vehicle.message_factory.command_long_encode( 0, 0, # target_system, target_component mavutil.mavlink.MAV_CMD_CONDITION_YAW, #command 0, #confirmation 1, # param 1, yaw in degrees 0, # param 2, yaw speed deg/s heading, # param 3, direction -1 ccw, 1 cw 1, # param 4, relative offset 1, absolute angle 0 0, 0, 0) # param 5 ~ 7 not used

send command to vehicle

vehicle.send_mavlink(msg)
vehicle.flush()

-- Connect to the drone

vehicle = connect('udp:127.0.0.1:14550', wait_ready=True)

cap = cv2.VideoCapture(0) width=cap.get(cv2.CAP_PROP_FRAME_WIDTH) height=cap.get(cv2.CAP_PROP_FRAME_HEIGHT) tracker = cv2.legacy.TrackerMOSSE_create() Hide quoted text

def drawBox(img, box, distance): x, y, w, h = int(box[0]), int(box[1]), int(box[2]), int(box[3]) cv2.rectangle(img, (x, y), (x + w, y + h), (255, 0, 255), 2, cv2.LINE_AA) cv2.putText(img, "Estimated distance: {}cm".format(int(distance)), (75, 75), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 255), 2) cv2.line(img, (int(width/2), int(height/2)), (int(x+w/2), int((y+h/2))), (255,255,255), 1)

while True: _, img = cap.read()

Convert BGR to HSV

hsv = cv2.cvtColor(img, cv2.COLOR_BGR2HSV)
# define range of blue color in HSV
lower_blue = np.array([100,50,50])
upper_blue = np.array([130,255,255])
# Threshold the HSV image to get only blue colors
mask = cv2.inRange (hsv, lower_blue, upper_blue)
contours = cv2.findContours(mask.copy(),
                          cv2.RETR_EXTERNAL,
                          cv2.CHAIN_APPROX_SIMPLE)[-2]

if len(contours)>0:
    contour = max(contours, key=cv2.contourArea)
    x, y, w, h = cv2.boundingRect(contour)
    cv2.rectangle(img,(x,y),(x+w, y+h),(0,255,0),2)

    #-- Initialize the tracker
    tracker.init(img, (x, y, w, h))

    distance = int(474 * (80 / w)) if w != 0 else 0
    #print(distance)

#-- Draw the bounding box and estimated distance
    drawBox(img, [x, y, w, h], distance)
    errorx = abs((x + w/2)- width)
    errory = abs((y + h/2) - height)
    print(distance, errorx, errory)

    if distance > 120:
        send_local_ned_velocity(1, 0, 0)

    if distance < 100:
        send_local_ned_velocity(-1, 0, 0)

    #if int(errorx > 700):
        #condition_yaw(1)
    #if int(errorx < 500):
        #condition_yaw(-1)

    #if int(errory > 200):
    #    send_local_ned_velocity(0, 0, -1)
    #if int(errory < 100):
    #    send_local_ned_velocity(0, 0, 1)

    #-- Write video frame and display

    cv2.imshow('Image', img)

    k = cv2.waitKey(5)
    if k == 27:
        break

cap.release() cv2.destroyAllWindows()

Running on jetson and intel cpu should not affect the functionality of code or behaviour

Velocity Function: Ensure you're using the correct frame for your velocity commands. You might want to switch from MAV_FRAME_BODY_OFFSET_NED to MAV_FRAME_LOCAL_NED to see if that improves the response.

Condition for Movement: The transitions between moving forward and backward may be too abrupt. If the distance fluctuates around the thresholds (100-120), the drone could be rapidly switching between velocities. Adding a deadband around these checks could help stabilize the movement.

Yaw Control: If you're adjusting yaw at the same time as changing X velocity, it might lead to circular motion. Make sure yaw adjustments don’t conflict with velocity commands.

Error Calculation: Review how you're calculating errorx and errory. Large, unadjusted errors could result in significant changes in velocity commands.

Loop Consistency: Consider introducing a small delay (time.sleep()) in your loop to control command frequency, which might help stabilize behavior.

Tracking Initialization: Ensure that the tracker is initialized correctly. An unstable bounding box can lead to erratic movements.

punkypankaj / Autonomous-Takeoff-and-Land-Python-Script