Skynet is Tracking 16 Bogeys

erco · 2011-08-31 09:24

Incoming!

http://www.youtube.com/watch?v=dXiubArqPQk&feature=digest_wed

The Terminator will surely rise from ExDiv's workshop.

Martin_H · 2011-08-31 19:59

Amazing and it also puts my two machine vision projects to shame.

ExDxV · 2011-10-04 11:01

Machine that can see!

Route training and navigation by map:

See also: AVM Navigator help page, Using of AVM plugin in RoboRealm

Notification:

*What your robot see and how it affect to navigation?

In our case the robot uses only limited sequence of images from camera for
navigation. The AVM Navigator application just tries to recognize images so
that be able to understand robot location. And if you show in route training
process of some image sequence (just lead your robot in "Marker mode")
then in automatic navigation mode the robot should have possibility to see
the same sequence of images without anyone changes. Otherwise AVM Navigator
will not be able to recognize images and it will cause to fail in
localization and navigation.

Martin_H · 2011-10-04 11:27

That was really cool, thanks for posting that.

ExDxV · 2011-10-18 01:28

It's test of new algorithm for AVM Navigator v0.7.3.

First in video the robot has received command: "go to the checkpoint" and when robot arrived to the checkpoint then I brought robot back (several times) in different position of learned route. When robot noticed the changes then it indicated that robot was displaced because any commands were not given by robot to his motors however changes were seen in the input image.

Then robot started looking around and localized his current position. Further the robot just calculated path from current position to the checkpoint and went there (and so forth).

ExDxV · 2012-04-07 07:33

Presentation of Digital Video Recording system (DVR)

Ttailspin · 2012-04-07 07:54

I am just a hobbiest... So the only comment I can make is.. WOW!, very cool!.

-Tommy

ExDxV · 2012-04-21 12:24

It is a testing of the enhanced tracking algorithm that takes into consideration variable servo speed:

See VBScript program and diagram below for more details:

' Get turret control variables
flt_base = 10000

rate_v  = GetVariable("RATE_V")/flt_base
turret_v  = GetVariable("TURRET_V")/flt_base
turret_Sv = GetVariable("TURRET_V_SPEED")

rate_h  = GetVariable("RATE_H")/flt_base
turret_h  = GetVariable("TURRET_H")/flt_base
turret_Sh = GetVariable("TURRET_H_SPEED")
turret_f  = GetVariable("TURRET_FIRE")
step_counter = GetVariable("STEP_COUNTER")

dX = 0
dY = 0

status = ""
turret_v_initial = -80

nvObjectsTotal = GetVariable("NV_OBJECTS_TOTAL")

if nvObjectsTotal>0 then   ' If any object was found

	' Get image size
	img_w = GetVariable("IMAGE_WIDTH")
	img_h = GetVariable("IMAGE_HEIGHT")
	
	' Get array variables of recognized objects
	nvArrObjRectX = GetArrayVariable("NV_ARR_OBJ_RECT_X")
	nvArrObjRectY = GetArrayVariable("NV_ARR_OBJ_RECT_Y")
	nvArrObjRectW = GetArrayVariable("NV_ARR_OBJ_RECT_W")
	nvArrObjRectH = GetArrayVariable("NV_ARR_OBJ_RECT_H")

	' Get center coordinates of first object from array
	obj_x = nvArrObjRectX(0) + nvArrObjRectW(0)/2
	obj_y = nvArrObjRectY(0) - nvArrObjRectH(0)/2
	
	' Get difference between object and screen centers
	dX = img_w/2 - obj_x
	dY = img_h/2 - obj_y
	
	dXr = 1 - abs(dX*4/img_w)
	if dXr < 0 then dXr = 0
	
	dYr = 1 - abs(dY*4/img_h)
	if dYr < 0 then dYr = 0
	
	turret_min = -100
	turret_max = 100
	reaction   = 7
	speed_min  = 1
	speed_max  = 100
	filtering  = 0.7
	decay      = 0.1
	threshold  = round(img_w*0.03)

	sRateH = exp(-dXr*reaction)
	sRateV = exp(-dYr*reaction)
	
	rate_h = rate_h + (sRateH - rate_h)*filtering
	rate_v = rate_v + (sRateV - rate_v)*filtering
	
	turret_Sh = round(speed_min + rate_h*(speed_max - speed_min))
	turret_Sv = round(speed_min + rate_v*(speed_max - speed_min))
	
	delta_h = (img_w/8)*rate_h
	delta_v = (img_h/8)*rate_v

  if step_counter =< 0 then
		step_counter = round(exp(-(dXr*dYr)*reaction*0.7)*15)
		
		if dX > threshold then
			' The object is at left side
			turret_h = turret_h - delta_h
		
			if turret_h < turret_min then turret_h = turret_min
		end if

		if dX < -threshold then
			' The object is at right side
			turret_h = turret_h + delta_h
		
			if turret_h > turret_max then turret_h = turret_max
		end if
	
		if dY > threshold then
			' The object is at the bottom
			turret_v = turret_v - delta_v
		
			if turret_v < turret_min then turret_v = turret_min
		end if
	
		if dY < -threshold then
			' The object is at the top
			turret_v = turret_v + delta_v
		
			if turret_v > turret_max then turret_v = turret_max
		end if
	else
		step_counter = step_counter - 1
	end if
		
	' Is the target locked?
	if dX < threshold and dX > -threshold and dY < threshold and dY > -threshold then
		status = "Target is locked"
		turret_f = 1
	else
		status = "Tracking"
		turret_f = 0
	end if
else
	' Back to the center if object is lost
	if turret_h > 0 then turret_h = turret_h - 1
	if turret_h < 0 then turret_h = turret_h + 1
	if turret_v > turret_v_initial then turret_v = turret_v - 1
	if turret_v < turret_v_initial then turret_v = turret_v + 1
	
	turret_Sh = speed_min
	turret_Sv = speed_min
	
	rate_h = rate_h - rate_h*decay
	rate_v = rate_v - rate_v*decay
	
	turret_f = 0
end if

' Set turret control variables
SetVariable "RATE_V", rate_v*flt_base
SetVariable "TURRET_V", turret_v*flt_base
SetVariable "TURRET_V_CONTROL", round(turret_v)
SetVariable "TURRET_V_SPEED", turret_Sv
SetVariable "RATE_H", rate_h*flt_base
SetVariable "TURRET_H", turret_h*flt_base
SetVariable "TURRET_H_CONTROL", round(turret_h)
SetVariable "TURRET_H_SPEED", turret_Sh
SetVariable "TURRET_FIRE", turret_f
SetVariable "STEP_COUNTER", step_counter
SetVariable "DELTA_X", dX
SetVariable "DELTA_Y", dY
SetVariable "TURRET_STATUS", status

attachment.php?attachmentid=3966&d=1335029045&thumb=1

plingboot wrote:

I've had a 'fiddle' with AVM navigator and managed to teach it a few objects/faces, but not the first idea how to turn that into tracking commands

It is easy. You should just use "Object recognition mode" in AVM Navigator module.
First clear the AVM search tree by click on button "Set key image size(New)" and further press "Yes".

Now you can train AVM on some faces like in video below:

Face training demo

When training will be done you should use variables that described below for your VBScript program:

NV_OBJECTS_TOTAL - total number of recognized objects
NV_ARR_OBJ_RECT_X - left-top corner X coordinate of recognized object
NV_ARR_OBJ_RECT_Y - left-top corner Y coordinate of recognized object
NV_ARR_OBJ_RECT_W - width of recognized object
NV_ARR_OBJ_RECT_H - height of recognized object

As example you can use these VBScript programs that was published in this topics:
http://www.roborealm.com/forum/index.php?thread_id=3881#
http://forums.trossenrobotics.com/showthread.php?4764-Using-of-AVM-plugin-in-RoboRealm&p=48865#post48865

erco · 2012-04-21 13:51

@ExDxV: Thanks for another amazing demo video. Your "Terminator" sunglasses only prove my prediction that Skynet will arise from the excellent work that you are doing!

ExDxV · 2012-04-21 22:00

I just try to create machine that can see real world and I think that this solution is pretty possible because Mother Nature already made it a long time ago

ExDxV · 2012-06-24 12:35

This is test of Human tracking by AVM Navigator in "Object recognition" mode with "Learn from motion" option.

erco · 2012-06-24 19:25

Beautiful! So do wear all-white exclusively, or does help the software track you?

ExDxV · 2012-06-24 22:25

The source information for tracking is motion history between frames.
So if I would clothe the other garments that have contraster texture it could work just better

Skynet is Tracking 16 Bogeys

Comments