Skip to content

Instantly share code, notes, and snippets.

@TimMcCauley
Created December 10, 2020 07:41
Show Gist options
  • Save TimMcCauley/5bd61b968849ebca4974b636e16f4fa9 to your computer and use it in GitHub Desktop.
Save TimMcCauley/5bd61b968849ebca4974b636e16f4fa9 to your computer and use it in GitHub Desktop.
This lua file is used in OSRM v5.23.0 and makes use of polygons provided to manipulate the edge forward and backward speed. Look for "customized speed manipulation" in the lua script.
api_version = 4
Set = require('lib/set')
Sequence = require('lib/sequence')
Handlers = require("lib/way_handlers")
Relations = require("lib/relations")
find_access_tag = require("lib/access").find_access_tag
limit = require("lib/maxspeed").limit
Utils = require("lib/utils")
Measure = require("lib/measure")
function setup()
return {
properties = {
max_speed_for_map_matching = 180/3.6, -- 180kmph -> m/s
-- For routing based on duration, but weighted for preferring certain roads
weight_name = 'routability',
-- For shortest duration without penalties for accessibility
-- weight_name = 'duration',
-- For shortest distance without penalties for accessibility
-- weight_name = 'distance',
process_call_tagless_node = false,
u_turn_penalty = 20,
continue_straight_at_waypoint = true,
use_turn_restrictions = true,
left_hand_driving = false,
traffic_light_penalty = 2,
},
default_mode = mode.driving,
default_speed = 10,
oneway_handling = true,
side_road_multiplier = 0.8,
turn_penalty = 7.5,
speed_reduction = 0.8,
turn_bias = 1.075,
cardinal_directions = false,
-- Size of the vehicle, to be limited by physical restriction of the way
vehicle_height = 2.0, -- in meters, 2.0m is the height slightly above biggest SUVs
vehicle_width = 1.9, -- in meters, ways with narrow tag are considered narrower than 2.2m
-- Size of the vehicle, to be limited mostly by legal restriction of the way
vehicle_length = 4.8, -- in meters, 4.8m is the length of large or family car
vehicle_weight = 2000, -- in kilograms
-- a list of suffixes to suppress in name change instructions. The suffixes also include common substrings of each other
suffix_list = {
'N', 'NE', 'E', 'SE', 'S', 'SW', 'W', 'NW', 'North', 'South', 'West', 'East', 'Nor', 'Sou', 'We', 'Ea'
},
barrier_whitelist = Set {
'cattle_grid',
'border_control',
'toll_booth',
'sally_port',
'gate',
'lift_gate',
'no',
'entrance',
'height_restrictor',
'arch'
},
access_tag_whitelist = Set {
'yes',
'motorcar',
'motor_vehicle',
'vehicle',
'permissive',
'designated',
'hov'
},
access_tag_blacklist = Set {
'no',
'agricultural',
'forestry',
'emergency',
'psv',
'customers',
'private',
'delivery',
'destination'
},
-- tags disallow access to in combination with highway=service
service_access_tag_blacklist = Set {
'private'
},
restricted_access_tag_list = Set {
'private',
'delivery',
'destination',
'customers',
},
access_tags_hierarchy = Sequence {
'motorcar',
'motor_vehicle',
'vehicle',
'access'
},
service_tag_forbidden = Set {
'emergency_access'
},
restrictions = Sequence {
'motorcar',
'motor_vehicle',
'vehicle'
},
classes = Sequence {
'toll', 'motorway', 'ferry', 'restricted', 'tunnel'
},
-- classes to support for exclude flags
excludable = Sequence {
Set {'toll'},
Set {'motorway'},
Set {'ferry'}
},
avoid = Set {
'area',
-- 'toll', -- uncomment this to avoid tolls
'reversible',
'impassable',
'hov_lanes',
'steps',
'construction',
'proposed'
},
speeds = Sequence {
highway = {
motorway = 90,
motorway_link = 45,
trunk = 85,
trunk_link = 40,
primary = 65,
primary_link = 30,
secondary = 55,
secondary_link = 25,
tertiary = 40,
tertiary_link = 20,
unclassified = 25,
residential = 25,
living_street = 10,
service = 15
}
},
service_penalties = {
alley = 0.5,
parking = 0.5,
parking_aisle = 0.5,
driveway = 0.5,
["drive-through"] = 0.5,
["drive-thru"] = 0.5
},
restricted_highway_whitelist = Set {
'motorway',
'motorway_link',
'trunk',
'trunk_link',
'primary',
'primary_link',
'secondary',
'secondary_link',
'tertiary',
'tertiary_link',
'residential',
'living_street',
'unclassified',
'service'
},
construction_whitelist = Set {
'no',
'widening',
'minor',
},
route_speeds = {
ferry = 5,
shuttle_train = 10
},
bridge_speeds = {
movable = 5
},
-- surface/trackype/smoothness
-- values were estimated from looking at the photos at the relevant wiki pages
-- max speed for surfaces
surface_speeds = {
asphalt = nil, -- nil mean no limit. removing the line has the same effect
concrete = nil,
["concrete:plates"] = nil,
["concrete:lanes"] = nil,
paved = nil,
cement = 80,
compacted = 80,
fine_gravel = 80,
paving_stones = 60,
metal = 60,
bricks = 60,
grass = 40,
wood = 40,
sett = 40,
grass_paver = 40,
gravel = 40,
unpaved = 40,
ground = 40,
dirt = 40,
pebblestone = 40,
tartan = 40,
cobblestone = 30,
clay = 30,
earth = 20,
stone = 20,
rocky = 20,
sand = 20,
mud = 10
},
-- max speed for tracktypes
tracktype_speeds = {
grade1 = 60,
grade2 = 40,
grade3 = 30,
grade4 = 25,
grade5 = 20
},
-- max speed for smoothnesses
smoothness_speeds = {
intermediate = 80,
bad = 40,
very_bad = 20,
horrible = 10,
very_horrible = 5,
impassable = 0
},
-- http://wiki.openstreetmap.org/wiki/Speed_limits
maxspeed_table_default = {
urban = 50,
rural = 90,
trunk = 110,
motorway = 130
},
-- List only exceptions
maxspeed_table = {
["at:rural"] = 100,
["at:trunk"] = 100,
["be:motorway"] = 120,
["be-vlg:rural"] = 70,
["by:urban"] = 60,
["by:motorway"] = 110,
["ch:rural"] = 80,
["ch:trunk"] = 100,
["ch:motorway"] = 120,
["cz:trunk"] = 0,
["cz:motorway"] = 0,
["de:living_street"] = 7,
["de:rural"] = 100,
["de:motorway"] = 0,
["dk:rural"] = 80,
["fr:rural"] = 80,
["gb:nsl_single"] = (60*1609)/1000,
["gb:nsl_dual"] = (70*1609)/1000,
["gb:motorway"] = (70*1609)/1000,
["nl:rural"] = 80,
["nl:trunk"] = 100,
['no:rural'] = 80,
['no:motorway'] = 110,
['pl:rural'] = 100,
['pl:trunk'] = 120,
['pl:motorway'] = 140,
["ro:trunk"] = 100,
["ru:living_street"] = 20,
["ru:urban"] = 60,
["ru:motorway"] = 110,
["uk:nsl_single"] = (60*1609)/1000,
["uk:nsl_dual"] = (70*1609)/1000,
["uk:motorway"] = (70*1609)/1000,
['za:urban'] = 60,
['za:rural'] = 100,
["none"] = 140
},
relation_types = Sequence {
"route"
},
-- classify highway tags when necessary for turn weights
highway_turn_classification = {
},
-- classify access tags when necessary for turn weights
access_turn_classification = {
}
}
end
function process_node(profile, node, result, relations)
-- parse access and barrier tags
local access = find_access_tag(node, profile.access_tags_hierarchy)
if access then
if profile.access_tag_blacklist[access] and not profile.restricted_access_tag_list[access] then
result.barrier = true
end
else
local barrier = node:get_value_by_key("barrier")
if barrier then
-- check height restriction barriers
local restricted_by_height = false
if barrier == 'height_restrictor' then
local maxheight = Measure.get_max_height(node:get_value_by_key("maxheight"), node)
restricted_by_height = maxheight and maxheight < profile.vehicle_height
end
-- make an exception for rising bollard barriers
local bollard = node:get_value_by_key("bollard")
local rising_bollard = bollard and "rising" == bollard
if not profile.barrier_whitelist[barrier] and not rising_bollard or restricted_by_height then
result.barrier = true
end
end
end
-- check if node is a traffic light
local tag = node:get_value_by_key("highway")
if "traffic_signals" == tag then
result.traffic_lights = true
end
end
function process_way(profile, way, result, relations)
-- the intial filtering of ways based on presence of tags
-- affects processing times significantly, because all ways
-- have to be checked.
-- to increase performance, prefetching and intial tag check
-- is done in directly instead of via a handler.
-- in general we should try to abort as soon as
-- possible if the way is not routable, to avoid doing
-- unnecessary work. this implies we should check things that
-- commonly forbids access early, and handle edge cases later.
-- data table for storing intermediate values during processing
local data = {
-- prefetch tags
highway = way:get_value_by_key('highway'),
bridge = way:get_value_by_key('bridge'),
route = way:get_value_by_key('route')
}
-- perform an quick initial check and abort if the way is
-- obviously not routable.
-- highway or route tags must be in data table, bridge is optional
if (not data.highway or data.highway == '') and
(not data.route or data.route == '')
then
return
end
handlers = Sequence {
-- set the default mode for this profile. if can be changed later
-- in case it turns we're e.g. on a ferry
WayHandlers.default_mode,
-- check various tags that could indicate that the way is not
-- routable. this includes things like status=impassable,
-- toll=yes and oneway=reversible
WayHandlers.blocked_ways,
WayHandlers.avoid_ways,
WayHandlers.handle_height,
WayHandlers.handle_width,
WayHandlers.handle_length,
WayHandlers.handle_weight,
-- determine access status by checking our hierarchy of
-- access tags, e.g: motorcar, motor_vehicle, vehicle
WayHandlers.access,
-- check whether forward/backward directions are routable
WayHandlers.oneway,
-- check a road's destination
WayHandlers.destinations,
-- check whether we're using a special transport mode
WayHandlers.ferries,
WayHandlers.movables,
-- handle service road restrictions
WayHandlers.service,
-- handle hov
WayHandlers.hov,
-- compute speed taking into account way type, maxspeed tags, etc.
WayHandlers.speed,
WayHandlers.surface,
WayHandlers.maxspeed,
WayHandlers.penalties,
-- compute class labels
WayHandlers.classes,
-- handle turn lanes and road classification, used for guidance
WayHandlers.turn_lanes,
WayHandlers.classification,
-- handle various other flags
WayHandlers.roundabouts,
WayHandlers.startpoint,
WayHandlers.driving_side,
-- set name, ref and pronunciation
WayHandlers.names,
-- set weight properties of the way
WayHandlers.weights,
-- set classification of ways relevant for turns
WayHandlers.way_classification_for_turn
}
WayHandlers.run(profile, way, result, data, handlers, relations)
-- customized speed manipulation
forward_sp = result.forward_speed
backward_sp = result.backward_speed
factor = 1.0
if way:get_location_tag('speed_factor') then
factor = way:get_location_tag('speed_factor')
end
if forward_sp and forward_sp > 0 then
result.forward_speed = forward_sp * factor
end
if backward_sp and backward_sp > 0 then
result.backward_speed = backward_sp * factor
end
-- end customized speed manipulation
if profile.cardinal_directions then
Relations.process_way_refs(way, relations, result)
end
end
function process_turn(profile, turn)
-- Use a sigmoid function to return a penalty that maxes out at turn_penalty
-- over the space of 0-180 degrees. Values here were chosen by fitting
-- the function to some turn penalty samples from real driving.
local turn_penalty = profile.turn_penalty
local turn_bias = turn.is_left_hand_driving and 1. / profile.turn_bias or profile.turn_bias
if turn.has_traffic_light then
turn.duration = profile.properties.traffic_light_penalty
end
if turn.number_of_roads > 2 or turn.source_mode ~= turn.target_mode or turn.is_u_turn then
if turn.angle >= 0 then
turn.duration = turn.duration + turn_penalty / (1 + math.exp( -((13 / turn_bias) * turn.angle/180 - 6.5*turn_bias)))
else
turn.duration = turn.duration + turn_penalty / (1 + math.exp( -((13 * turn_bias) * -turn.angle/180 - 6.5/turn_bias)))
end
if turn.is_u_turn then
turn.duration = turn.duration + profile.properties.u_turn_penalty
end
end
-- for distance based routing we don't want to have penalties based on turn angle
if profile.properties.weight_name == 'distance' then
turn.weight = 0
else
turn.weight = turn.duration
end
if profile.properties.weight_name == 'routability' then
-- penalize turns from non-local access only segments onto local access only tags
if not turn.source_restricted and turn.target_restricted then
turn.weight = constants.max_turn_weight
end
end
end
return {
setup = setup,
process_way = process_way,
process_node = process_node,
process_turn = process_turn
}
@phcantis
Copy link

phcantis commented Jul 1, 2021

Thank you for your response Tim!

regarding question 1 - just to clarify - could london.geojson hold more than one single polygon? e.g. n areas that flood and overlap with street segments within london.

For example, a .json file like the following below, with 2 geometries [areas flooded], would work? (also I foresee that I will have many geometries, in the 100s or 1000s)

{"type":"FeatureCollection","crs":{"type":"name","properties":{"name":"EPSG:26918"}},"features":[{"type":"Feature","id":0,"geometry":{"type":"Polygon","coordinates":[[[599687.0990039676,4516071.7771109855],[599686.89171178814,4516072.4379095687],[599687.42897854745,4516072.6064011483],[599683.72309983696,4516084.4230810879],[599672.78987609863,4516080.9943190096],[599676.47075100779,4516069.2573376466],[599676.82331754465,4516069.3678350858],[599677.05551337055,4516068.6273371056],[599687.0990039676,4516071.7771109855]]]},"properties":{"FID":0,"new_speed":20}},{"type":"Feature","id":1,"geometry":{"type":"Polygon","coordinates":[[[599772.63732254691,4516095.9699044283],[599773.51327440294,4516096.2475355314],[599768.83976576838,4516110.9957737578],[599761.59993338655,4516108.7015918251],[599760.43618100707,4516108.3327977397],[599765.08936211385,4516093.6487146541],[599766.15767968807,4516093.9872267181],[599766.61037690716,4516092.5585861932],[599768.45592401538,4516093.1434100932],[599768.79713207553,4516092.0666250121],[599773.24732469139,4516093.4768790873],[599773.41089934821,4516093.528733613],[599772.63732254691,4516095.9699044283]]]},"properties":{"FID":1,"new_speed":0}}]}

@TimMcCauley
Copy link
Author

I don't think so, possibly a MultiPolygon will work but that then would use the same settings for all polygons within. Just give it a try, you will find eventually out :-)

@phcantis
Copy link

phcantis commented Jul 1, 2021

Thanks! In that case it wouldnt be that bad. I could just have two sequential multipolygon jsons, one with each new speeds according to the two different depths im considering.

I cant wait to dive in! Thank you for your help. I am 0 familiar with lua so if there is an editor or viewer I should seek, any link to a set up tutorial / osm lua files handling tutorial would be amazing. Will report back if I have results.

@TimMcCauley
Copy link
Author

Reg. the lua files, there are many flying around as gists and there is quite some information in the repo itself, you will have to browse around a bit. Printing within the lua file is usually my best friend. If Multipolygon does not work, you would also "script together" the CLI command if there are hundreds of location data arguments.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment