Move_basic tutorial / is my map ok?

[brunolfm]

@davecrawley today I was finally resume the testings on the robot! The robot stand still for the last 3weeks with no one touching him! But today when I press both the buttons to turn it on I hear a noise and saw a sparkle from the board! But the robot remained on! But I can’t do anything! Even connecting the monitor directly to the HDMI port the robot seems dead! But all the lights are normal! When I made an visual inspection I noticed that this resistor may have burned! But I can’t tell! Can you see in the picture? Later I will disassemble the main board to make a thorough inspection! This is bad I was so hyped to resume the testing and this happens

[davecrawley]

The part you highlight is the real time clock. They’ve never been known to fail - but there is always a first time. Also the good news is the robot should operate without it - however if it fails badly then it could short out the power and cause all kinds of problems.

Please do the following initial checks.

1. Make sure there are no stray metal parts (Screws, stray wires etc.) making contact with either the rear side of the connector to the sonar board or to the under-side of the sonar jack.

2. Turn on both the motor and main power - see if the wheels are mostly locked in place and see if they resist movement. When you try to turn the wheels with you hands you should feel resistance and the wheels should move back mostly to to the original position on their own.

3. Check to see if the LED lights above the switch board are on. All 5 lights should be on the top light should blink briefly every 6 seconds.

Let me know how you get on and we can help you. If the above are fine then.

4. Power on the robot and see if you can log in to to the robot - I am guessing that the answer is no.

5. If you can’t log in to the robot, disconnect the Pi and plug it in to a separate power supply (e.g. a phone charger) and try to see if you can connect to the raspberry pi by itself. If you can’t the Raspberry pi is probably dead. We’ve seen the raspberry pi spontaneously fail a few times, the good news is they are easy to replace.

6. If the raspberry pi has failed we’ll need to replace it - let me know if you’d like me to order a new one for you.

7. If the raspberry pi has not failed but you aren’t able to log in when its connected to the robot then the fault is on the motor board, and we’ll have to repair it. email dc@ubiquityrobotics.com and we’ll arrange a return. You can try removing the damaged chip yourself if you like - if that’s the only failure the robot will work - however it will not keep an accurate clock. This might not affect you, though it can affect some other operations.

[brunolfm]

1. Didnt notice anything irregular on the sonar board! didnt remove-it tough! just minimal visual check-up!
2. Yes. The robot locks the movements of the wheels and if i try to move it, he imediatly forces back to the original position!
3. Correct, all the 5 lights turned on and the top one blinks every some seconds!
4. No, ive connected the hdmi directly to the Pi but didnt see image on the monitor!
5. I tryed to do that to no avail! the Pi seems dead! i connected to different outlets and different chargers and the Pi just turns on the red light nothing more! the boot didnt start! It seems to me that this is the problem. @davecrawley can i buy this one to replace it? https://www.amazon.co.uk/Raspberry-Pi-Model-64-Bit-Processor/dp/B07BDR5PDW/ref=sr_1_1_sspa?ie=UTF8&qid=1543794062&sr=8-1-spons&keywords=raspberry+pi+3+model+b%2B&psc=1

[davecrawley]

Yes it appears you have a dead raspberry pi - but it appears your motor controller is functioning correctly. Yes if you get a new Raspberry pi 3B+ and replace it that would be the first thing to try.

[brunolfm]

Ordered! I will receive it on 5 December! And then I will tell you the results! Thanks for the support!

[davecrawley]

It is possible that there is a board defect that is causing the failure of the pi. However usually its the Raspberry Pi 3B+. Either way we’ll reimburse you for the burned pis.

[brunolfm]

Please @davecrawley you dont have to pay for this! The support you guys give to me is more than enough!

Check the email i´ve send to you please! I need that asap!

[brunolfm]

Hello again everyone!

Im returning to the tests now that i have the robot in 100% condition as @davecrawley help me with the sonar board! Many thanks to you again!

I´ve put the robot doing an circuit and everything was going well when at the end of almost 12 straight hours the robot has stoped moving because it ran out of space… i think!

This was the error on the terminal:

-----IOError: [Errno 28] No space left on device-----

And on the the terminal that i was ruinning the: roslaunch magni_nav aruco.launch the last “Got image” was at 125.000k! Maybe this is too many… i dont know!!

some help would be appreciated!

[davecrawley]

I am glad to hear that you had this running for 12 hours. A couple of clarifying questions.

1. Are you gathering bag files or other data?
2. Are you using the standard SD card that we provided with the robot? If you aren’t
   • What is the size of the SD card that you are using?
   • Is it a branded device (e.g. Sandisk or other)? We’ve had a lot of trouble with fake SD cards that are everywhere - they appear to be a certain size but aren’t in fact that size.
   • What is the speed of the device you are using (e.g. class 10, HC1, HC3) etc.

[brunolfm]

@davecrawley the pi on the robot was randomly creating aproximately 100log files per second, and that was the case, i manage to fix that! The sd card is genuine 32gb kingston class 10 hc! I will try now again now that i believe all the problems are solved!

@alexis i know ive promised a time-lapse but i was not able to do it in the time being, i will let a video here with the robot doing a complete path on my garage, basicaly the magni do this in 12+ straight hours. On my final project the robot will do the exact same thing, but on a bigger space, and it will have “checkpoints” to precise localization every 10meters, and “stop points” where it will stop for 2 minutes for loading purposes!

My objective is to put 7/8 robots in line, every one following the robot in front of him, like a train!

Im very optimistic, and i believe i have the project 90% complete! Right now im waiting for the final warehouse to be finished to start testing on the production scenario!

BTW this is not the final cabinet! The final cabinet is gona be made in “T-Frame Aluminium”, again, this is for tests only

Video:

[davecrawley]

Awesome!!

What was the fix to the problem of running out of space - we will incorporate this in to the standard image and support it.

When you set up your fiducial field do you have a sense for the speed you use when mapping. It seems that you get much better results if you drive on the slower end - is that what you did?

[brunolfm]

@davecrawley the problem of the device running out of space its related to a bad script i have running on the magni´s pi! Its not ubiquity vm image related!

When i have no map, i´ve used the robot commander android app to generate the first map, with a couples of times passing by all the fiducials with 0.3/0.3 on the linear/angular speed seted-up on the app!

Then i´ve put the robot in automatic-loop mode passing for all the fiducials using the @alexis script at the same velocity for aproximately 10hours/day! Then when i believe i have a good map i put the map in “read_only” mode and its done!

@davecrawley can you help me with this question? i have an emergency stop button i would like to attach to the magni board in order to shut-him down if an emergency occur! Can you or someone guide me trough that?

[davecrawley]

Yes there are specific pins you need to attach it to. You need to attach it between

EN_Motor
and any 24V pin on
P601

To find P601 and see which pins are EN_Motor check out this document:

Magni Pinout

Although before you do this - check to see if you get E-Stop functionality on your Magni currently. The Red button on your power board should stop the motors from driving - although some people from the first batch had trouble with this functionality working.

You’ll need to remove the switch board and put your big red button in place of the red switch on the switch board.

[brunolfm]

Magni hands of surgeon!

[alexis]

well done!

[davecrawley]

I love it!! All the work that we did on smoothing out the motions of the robot is paid off with this!

[mjstn2011]

Per a big ESTOP switch: We have in development now a version of the ‘switch board’ that will support a user having a cable to go to an off-board ESTOP switch and/or an offboard On/Off switch. I attach a picture of the first test run on this new board.

[lorne]

Hello. @alexis
I was going to copy your goal sender but I can’t seem to get the cmakelist and package xml right.
would you mind sharing those as well?

start the thing with roslaunch magni_goal_sender go.launch

To get the coordinates, I put the yellow tape on the floor. then used rosrun tf tf_echo map base_link while teleop the magni to get to the corresponding orientation and built the yaml file accordingly

it’s very crude but it should get you up and running if you struggle to send goals.

ubuntu@magni02:~$ cat ~/catkin_ws/src/magni_goal_sender/param/goals.yaml
# x , y, yaw
# units; x/y => meters  |  yeaw => degrees
# tuse teleop to move magni, then use the following commande to see current pose
# rosrun tf tf_echo map base_link

goals:
 - 1.163, 0.451, 23.525
 - 2.296, 0.570, -38.058
 - 2.457, -0.384, -131.676
 - 1.224, -0.611, -173.067
 - 0.153, 0.158, 143.175
 - -1.035, 0.215, 179.176
 - -1.053, -1.386, -67.361
 - 0.0, 0.0, 0.0

ubuntu@magni02:~$ cat ~/catkin_ws/src/magni_goal_sender/launch/go.launch
<?xml version="1.0" encoding="UTF-8"?>
<launch>
  <!-- send goals -->
  <node name="goal_sender" pkg="magni_goal_sender" type="goal_sender.py" output="screen">
    <param name="param_filename" value="$(find magni_goal_sender)/param/goals.yaml"/>
    <param name="rate" value="25"/>
  </node>
</launch>
ubuntu@magni02:~$ cat ~/catkin_ws/src/magni_goal_sender/scripts/
goal_sender.py  test.py
ubuntu@magni02:~$ cat ~/catkin_ws/src/magni_goal_sender/scripts/goal_sender.py
#!/usr/bin/env python

import rospy
import yaml

#from math import sin, cos, pi
from math import radians # https://docs.python.org/2/library/math.html
from tf import transformations # for euler to

import time

#ubuntu@magni02:~$ rostopic info /move_base_simple/goal
#Type: geometry_msgs/PoseStamped
from geometry_msgs.msg import PoseStamped

#ubuntu@magni02:~$ rostopic info /move_base/status
#Type: actionlib_msgs/GoalStatusArray
#
#Publishers:
# * /move_basic (http://magni02.local:34847/)
from actionlib_msgs.msg import GoalStatusArray



class GoalSender:

    def __init__(self):
        rospy.init_node("goal_sender")
        self.nodename = rospy.get_name()
        rospy.loginfo("%s started" % self.nodename)

        #### parameters #######
        self.param_filename = rospy.get_param('~param_filename','/home/ubuntu/catkin_ws/src/magni_goal_sender/param/goals.yaml')  # where to find the goal list in yaml format
        rospy.loginfo("self.param_filename: %s" % self.param_filename)
        self.rate = rospy.get_param('~rate',20)  # how often do we update in Hz
        rospy.loginfo("self.rate: %s" % str(self.rate))
        self.load_goals()
        self.status = 3 # waiting for a new goal

        #### Publishers ####
        #self.goal_pub = rospy.Publisher("/move_base_simple/goal", geometry_msgs/PoseStamped, queue_size=10)
        self.goal_pub = rospy.Publisher("/move_base_simple/goal", PoseStamped, queue_size=10)
        time.sleep(0.5) # there must be at least  a 0.25sec delay  between publisher creation and message published => for the magni to actually move !?

        #### Subscribers ####
        rospy.Subscriber("/move_base/status", GoalStatusArray, self.status_callback)



    def load_goals(self):
        content = ""
        with open('/home/ubuntu/catkin_ws/src/magni_goal_sender/param/goals.yaml', 'r') as content_file:
            content = content_file.read()
        #print(yaml.dump(yaml.load(content)))
        yml = yaml.load(content)
        self.goals = yml['goals']
        self.index = 0
        rospy.loginfo("done loading goals")

        #for i in range(0,len(self.goals)):
        #    goal = self.goals[i].split(',')
        #    print(self.to_pose_stamped(float(goal[0]), float(goal[1]), float(goal[2])))


    def spin(self):
        r = rospy.Rate(self.rate)
        while not rospy.is_shutdown():
            self.update()
            r.sleep()

    def update(self):
        # known statuses
        #status: 1 | text: "This goal has been accepted by the simple action server"]  | magni is executing the goal command
        #status: 3 | text: ""]  | the magni is waiting for a goal



        if self.status == 3:
            goal = self.goals[self.index].split(',')
            # sending goal using command line
            # ubuntu@magni02:~$ rostopic pub -1 /move_base_simple/goal geometry_msgs/PoseStamped '{header: {stamp: now, frame_id: "map"}, pose: {position: {x: 1.0, y: 0.0, z: 0.0}, orientation: {w: 1.0}}}'
            # publishing and latching message for 3.0 seconds
            # ubuntu@magni02:~$

            pose = self.to_pose_stamped(float(goal[0]), float(goal[1]), float(goal[2]))
            rospy.loginfo("Sending goal pose "+str(self.index)+" to /move_base_simple/goal")
            #rospy.loginfo(str(pose))
            self.goal_pub.publish(pose)
            self.index += 1
            if self.index >= len(self.goals):
                self.index = 0
            time.sleep(0.5) # give time for callback to be called


        elif self.status == 1:
            x = 0 # do nothing since magni is executing the move command
            # status_callback will move the current status back to '3' once the goal is reached
            #rospy.loginfo("waiting for goal to be reached")
        else:
            rospy.logwarn("unknown status: %s" % self.status)



    def to_pose_stamped(self,x,y,theta_in_degrees):
        pose = PoseStamped()
        pose.header.stamp = rospy.Time.now()
        pose.header.frame_id = "map"
        pose.pose.position.x = x
        pose.pose.position.y = y
        pose.pose.position.z = 0.0

        quaternion = transformations.quaternion_from_euler(0.0, 0.0, radians(theta_in_degrees))
        pose.pose.orientation.x = quaternion[0]
        pose.pose.orientation.y = quaternion[1]
        pose.pose.orientation.z = quaternion[2]
        pose.pose.orientation.w = quaternion[3]

        return pose
    # source of inspiration => https://github.com/xaxxontech/oculusprime_ros/blob/master/src/global_path_follower.py
    def status_callback(self, msg):
        #rospy.loginfo(msg)
        #rospy.loginfo(msg.status_list)
        if len(msg.status_list) == 0:
            return

        self.status = msg.status_list[len(msg.status_list)-1].status # get latest status


if __name__ == '__main__':
    """ main """
    goal_sender = GoalSender()
    goal_sender.spin()

[alexis]

hi @lorne

sorry for my late answer.
I think we lost all code regarding this. all you have is here. Just so you know, it was just a “fire and forget” piece of code. We just forgot about the magni and decided to wait for the plateform to mature before we would look into it again.

I beleive @brunolfm used this code above on magni in production environment in Portugal. Maybe he can help you?

good luck

[lorne]

Thanks. I’ve been doing a little testing with the new 2dnav, it seems nice.