Improve humanize walking variance #5147

mjmadsen · 2016-09-03T10:55:26Z

Account for bearing when adding in walking "sway"
We now only sway "left" and "right" as we head towards our destination

mention-bot · 2016-09-03T10:55:27Z

@mjmadsen, thanks for your PR! By analyzing the annotation information on this pull request, we identified @douglascamata, @sinap and @alexyaoyang to be potential reviewers

th3w4y · 2016-09-03T11:07:58Z

pokemongo_bot/human_behaviour.py

@@ -24,6 +24,9 @@ def random_lat_long_delta():
    # Return random value from [-.000025, .000025]. Since 364,000 feet is equivalent to one degree of latitude, this
    # should be 364,000 * .000025 = 9.1. So it returns between [-9.1, 9.1]
    return ((random() * 0.00001) - 0.000005) * 5
+
+def random_lat_long_delta2(bearing):
+    return ((random() * 0.00001) - 0.000005) * 5, ((random() * 0.00001) - 0.000005) * 5


How do you plan to use bearing in the calculation...?

If we're heading at 0/360, we only add variance to latitude. Do some more math between there to scale each respective to that. I'm not sure

Wait I got that backwards. If we're headed north (0/360), we'd change our longitude.

The Delta applies to the destination. So as long as the Delta is smaller than the (destination - origin), you will never walk backward.
And as it is now, it is small enough for that. So not sure you need to change anything.

th3w4y · 2016-09-03T11:10:26Z

pokemongo_bot/walkers/step_walker.py

@@ -106,3 +110,43 @@ def step(self):

    def _pythagorean(self, lat, lng):
        return sqrt((lat ** 2) + (lng ** 2))


If you import math

then this should become

math.sqrt

sohje · 2016-09-03T11:20:27Z

pokemongo_bot/walkers/step_walker.py

@@ -109,7 +109,7 @@ def step(self):
        #     alt, False)

    def _pythagorean(self, lat, lng):
-        return sqrt((lat ** 2) + (lng ** 2))
+        return math.sqrt((lat ** 2) + (lng ** 2))

    def calc_bearing(start_lat, start_lng, dest_lat, dest_lng):


Should it be static method? calc_bearing(self...) ? Or you forget @static decorator?
upd: i'm about calc_bearing()

@static # ?? def calc_bearing(start_lat, start_lng, dest_lat, dest_lng)

th3w4y · 2016-09-03T11:21:53Z

@mjmadsen I am afraid that the travis build is failing due to non-ASCII carracters

θ = atan2(sin(Δlong).cos(lat2),

sohje · 2016-09-03T11:24:05Z

pokemongo_bot/walkers/step_walker.py


    def step(self):
+        walk_sway = random_lat_long_delta2(self.bearing)


random_lat_long_delta2 import missed?

th3w4y · 2016-09-03T11:28:12Z

I don't think that adding a random value to only latitude or only longitude based on bearing adds much extra...

mjmadsen · 2016-09-03T11:30:20Z

@th3w4y this is for our discussion in #5140

We're adding in a bit of variance to simulate human walking. We want it to only move us "left" or "right".

sohje · 2016-09-03T11:32:16Z

pokemongo_bot/walkers/step_walker.py

-        return sqrt((lat ** 2) + (lng ** 2))
+        return math.sqrt((lat ** 2) + (lng ** 2))
+
+    def _calc_bearing(start_lat, start_lng, dest_lat, dest_lng):


make

def _calc_bearing(self, start_lat, start_lng, dest_lat, dest_lng)

or make it static

@staticmethod def calc_bearing(start_lat, start_lng, dest_lat, dest_lng) self.calc_bearing(self.initLat, self.initLng, self.dLat, self.dLng) #self.calc_bearing(self.initLat, self.initLng, self.dLat, self.dLng)

sohje · 2016-09-03T11:37:39Z

Works for me.

th3w4y · 2016-09-03T11:43:51Z

@sohje @mjmadsen I agreed with the idea... but the calculation doesn't seem to do that yet...

it doesn't even use bearing in the calculation.. thus it did not change anything yet...

sohje · 2016-09-03T11:45:24Z

pokemongo_bot/walkers/step_walker.py

@@ -4,7 +4,7 @@

 from random import uniform
 from pokemongo_bot.cell_workers.utils import distance
-from pokemongo_bot.human_behaviour import random_lat_long_delta2, sleep, random_alt_delta
+from pokemongo_bot.human_behaviour import random_lat_long_delta, random_lat_long_delta2, sleep, random_alt_delta


Any usage of random_lat_long_delta?

The test files use it still. I'll update those too if I can get this working :)

Actually the CI still fails because of how the test files are setup.

mjmadsen · 2016-09-03T11:45:27Z

@th3w4y Yeah, I still need to figure out the important part :)

sohje · 2016-09-03T11:45:57Z

@th3w4y @mjmadsen Just to make sure bot not crashing ;]

BreezeRo · 2016-09-03T13:05:38Z

looks good 💃 might have to edit pokemongo_bot/test/step_walker_test.py to pass CI

th3w4y · 2016-09-03T13:17:11Z

My proposal:

We only specify an effect in degrees from the bearing.. where we would want to be at

IIn [209]: from geopy.distance import VincentyDistance

In [210]: 

In [210]: from geopy import Point
In [225]: def get_next_pos(lat, lon, bearing, speed, offset_angle):
     ...:     origin = Point(lat, lon)
     ...:     lat, lon, _ = VincentyDistance(kilometers=speed*1e-3).destination(origin, bearing+random.randrange(-offset_angle, offset_angle))
     ...:     return lat, lon
     ...:     
     ...:     
     ...: 

In [226]: 

In [226]:  get_next_pos(ex_orig[0], ex_orig[1], 180, 3, 5)
Out[226]: (47.170610881050735, 8.516737739683606)

In [227]:  get_next_pos(ex_orig[0], ex_orig[1], 180, 3, 5)
Out[227]: (47.170610819427246, 8.516741190604915)

In [228]:  get_next_pos(ex_orig[0], ex_orig[1], 180, 3, 5)
Out[228]: (47.170610831755695, 8.516741880999465)

In [229]:

th3w4y · 2016-09-03T13:19:07Z

This above obviously can replace the whole pythagorean.. magnitude.. the big convoluted part in StepWalker

th3w4y · 2016-09-03T17:24:57Z

@sohje one explanation would be:

pokestops are too close... google maps does not have walking directions in between them

but your picture shows that you were automatically fallback to classic behaviour... waking through buildings

to confirm that 💯 % do you have the coordinates of those pokestops from your fort files?

mjmadsen · 2016-09-03T17:25:49Z

I'm running through all the walkers. Might not finish before I go to bed.

@anakin5 does this seem to give your poly the results you're looking for? If we seal this portion down, we don't have to mess with it for a long ass time!

sohje · 2016-09-03T17:27:29Z

@th3w4y according to google maps there are 1-3 routes between pokestops on that coords.

handle special values for speed

th3w4y · 2016-09-03T17:27:51Z

@mjmadsen one more commit:

https://github.com/mjmadsen/PokemonGo-Bot/pull/2
To make the function complete...(i don't want to see divisions by zero of walking of planet)

th3w4y · 2016-09-03T17:28:45Z

@mjmadsen i know that we'll not see 0.0 or infinite but it make the function complete and protected from ZeroDivizion

th3w4y · 2016-09-03T17:32:07Z

@mjmadsen I found a bug...

bearing should be towards the final destination to towards something calculated...

https://github.com/mjmadsen/PokemonGo-Bot/blob/13819891d3d0df4ec91fbe4bfeefcb7d64a1fbab/pokemongo_bot/walkers/step_walker.py#L58

th3w4y · 2016-09-03T17:33:35Z

This is incorrect

self.bearing = self._calc_bearing(self.initLat, self.initLng, self.dLat, self.dLng)

sohje · 2016-09-03T17:34:01Z

still moving straight ;]

[2016-09-03 20:31:12] [MoveToFort] Moving towards pokestop XX - 0.13km
[2016-09-03 20:31:17] [MoveToFort] Moving towards pokestop  XX - 0.11km
[2016-09-03 20:31:23] [MoveToFort] Moving towards pokestop XX - 0.10km
[2016-09-03 20:31:29] [MoveToFort] Moving towards pokestop  XX - 0.10km
[2016-09-03 20:31:35] [MoveToFort] Moving towards pokestop  XX - 0.11km
[2016-09-03 20:31:42] [MoveToFort] towards pokestop  XX - 0.12km

th3w4y · 2016-09-03T17:34:25Z

Should be:

self.bearing = self._calc_bearing(self.initLat, self.initLng, self.destLat, self.destLng)

sohje · 2016-09-03T17:36:36Z

yeap, with destLat and destLng works fine.

th3w4y · 2016-09-03T17:36:50Z

@mjmadsen bearing calculation should use final destination https://github.com/mjmadsen/PokemonGo-Bot/pull/3

bearing calculation should use final destination

mjmadsen · 2016-09-03T17:39:19Z

@sohje This should be a very unnoticeable change to look like strafing while walking.

th3w4y · 2016-09-03T17:40:25Z

@sohje so is working now?

screenshot please :)

sohje · 2016-09-03T17:41:59Z

@th3w4y yeap, with bearing fix works fine. Wait a sec, i'll make pull - edited manually.

sohje · 2016-09-03T17:43:26Z

hmm, i dont see on the map any strafing ;[

mjmadsen · 2016-09-03T17:43:42Z

@sohje me too

https://s15.postimg.org/4extwtzjf/test4.jpg

  {
    "type": "SpinFort",
    "config": {
      "enabled": true,
      "spin_wait_min": 3,
      "spin_wait_max": 5,
      "daily_spin_limit": 1900
    }
  },
  {
    "type": "MoveToFort",
    "config": {
      "enabled": true,
      "lure_attraction": true,
      "lure_max_distance": 2000,
      "walker": "PolylineWalker",
      "log_interval": 5
    }
  }
],
"map_object_cache_time": 5,
"forts": {
  "avoid_circles": false,
  "max_circle_size": 6,
  "cache_recent_forts": true
},

Turned catching pokemon off for testing

sohje · 2016-09-03T17:46:31Z

Seems we have to implement some sort of speed acceleration and deceleration.

th3w4y · 2016-09-03T17:48:33Z

@sohje for the polyline... that would be hard...

sohje · 2016-09-03T17:49:02Z

@th3w4y yeap...

th3w4y · 2016-09-03T17:49:21Z

@sohje i take the challenge then !
i'll need to add a dictionary or something to the Polyline class and modify all the position calculations based on all previous speeds at each previous .get_pos() call.

but i think i can do that if i also refactor the Polyline to use headings and bearings calculation...

will take few days...

th3w4y · 2016-09-03T18:43:30Z

but this does not need to wait for that refactoring...

BreezeRo · 2016-09-03T19:42:16Z

Tried, testing and working on my end - can confirm speed variance issue mentioned by @sohje

I'm approving this for now, will check in a moment before merging

mjmadsen added 3 commits September 3, 2016 05:50

Add bearing parameter

28bb383

Add calc_bearing

a454450

Fix placement

2238376

mjmadsen added 3 commits September 3, 2016 05:59

Switch to another function

8c212ff

Update step_walker.py

a99e309

Fix import

2edd135

th3w4y reviewed Sep 3, 2016
View reviewed changes

Merge branch 'dev' into mjmadsen-walker-variance

83c6e7e

th3w4y reviewed Sep 3, 2016
View reviewed changes

Update step_walker.py

23bcca0

sohje reviewed Sep 3, 2016
View reviewed changes

Update step_walker.py

41cf339

Update step_walker.py

14beb6d

sohje reviewed Sep 3, 2016
View reviewed changes

Update step_walker.py

51ccc01

Update step_walker.py

3331d56

sohje reviewed Sep 3, 2016
View reviewed changes

BreezeRo added the Waiting on Changes label Sep 3, 2016

Merge pull request #2 from th3w4y/variance

1381989

handle special values for speed

bearing calculation should use final destination

df1b17f

Update step_walker.py

810ebc4

Merge pull request #3 from th3w4y/variance

744fec0

bearing calculation should use final destination

BreezeRo added the Approved label Sep 3, 2016

th3w4y mentioned this pull request Sep 4, 2016

Refactor Polyline to support speed variance. #5157

Merged

solderzzc merged commit 6a30984 into PokemonGoF:dev Sep 4, 2016

th3w4y mentioned this pull request Sep 4, 2016

Improve step_walker #5184

Merged

		@@ -106,3 +110,43 @@ def step(self):

		def _pythagorean(self, lat, lng):
		return sqrt((lat 2) + (lng 2))


		def step(self):
		walk_sway = random_lat_long_delta2(self.bearing)

Improve humanize walking variance #5147

Improve humanize walking variance #5147

Conversation

mjmadsen commented Sep 3, 2016 • edited

mention-bot commented Sep 3, 2016

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sohje Sep 3, 2016 • edited

Choose a reason for hiding this comment

th3w4y commented Sep 3, 2016 • edited

Choose a reason for hiding this comment

th3w4y commented Sep 3, 2016 • edited

mjmadsen commented Sep 3, 2016

sohje Sep 3, 2016 • edited

Choose a reason for hiding this comment

sohje commented Sep 3, 2016 • edited

th3w4y commented Sep 3, 2016 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mjmadsen commented Sep 3, 2016

sohje commented Sep 3, 2016

BreezeRo commented Sep 3, 2016

th3w4y commented Sep 3, 2016 • edited

th3w4y commented Sep 3, 2016

th3w4y commented Sep 3, 2016 • edited

mjmadsen commented Sep 3, 2016

sohje commented Sep 3, 2016

th3w4y commented Sep 3, 2016

th3w4y commented Sep 3, 2016

th3w4y commented Sep 3, 2016 • edited

th3w4y commented Sep 3, 2016

sohje commented Sep 3, 2016

th3w4y commented Sep 3, 2016

sohje commented Sep 3, 2016

th3w4y commented Sep 3, 2016

mjmadsen commented Sep 3, 2016

th3w4y commented Sep 3, 2016

sohje commented Sep 3, 2016

sohje commented Sep 3, 2016

mjmadsen commented Sep 3, 2016 • edited

sohje commented Sep 3, 2016

th3w4y commented Sep 3, 2016

sohje commented Sep 3, 2016

th3w4y commented Sep 3, 2016 • edited

th3w4y commented Sep 3, 2016

BreezeRo commented Sep 3, 2016

mjmadsen commented Sep 3, 2016 •

edited

sohje Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited

sohje Sep 3, 2016 •

edited

sohje commented Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited

mjmadsen commented Sep 3, 2016 •

edited

th3w4y commented Sep 3, 2016 •

edited