add gaussian noise parameter

pfaedle.cfg

@@ -386,9 +386,6 @@ routing_station_unmatched_penalty: 0.3
 # If the platform does not match, add this penalty
 routing_platform_unmatched_penalty: 0.1
 
-# If the station name does not match, add this penalty
-routing_station_unmatched_penalty: 0.3
-
 # Max angle that should be counted as a full turn
 routing_full_turn_angle:			100
 
@@ -803,9 +800,6 @@ routing_full_turn_penalty:		120    # 2 minutes
 # Penalty added to non-station placements
 routing_non_station_penalty: 0.5
 
-# If the station name does not match, add this penalty
-routing_station_unmatched_penalty: 0.3
-
 # Max angle that should be counted as a full turn
 routing_full_turn_angle:			20
 

src/pfaedle/config/ConfigReader.cpp

@@ -136,6 +136,7 @@ void ConfigReader::read(Config* cfg, int argc, char** argv) {
                          {"write-colors", no_argument, 0, 13},
                          {"stats", no_argument, 0, 14},
                          {"no-hop-cache", no_argument, 0, 15},
+                         {"gaussian-noise", required_argument, 0, 16},
                          {0, 0, 0, 0}};
 
   char c;
@@ -205,6 +206,9 @@ void ConfigReader::read(Config* cfg, int argc, char** argv) {
       case 15:
         cfg->noHopCache = true;
         break;
+      case 16:
+        cfg->gaussianNoise = atof(optarg);
+        break;
       case 'v':
         std::cout << "pfaedle " << VERSION_FULL << std::endl;
         exit(0);

src/pfaedle/config/PfaedleConfig.h

@@ -35,7 +35,8 @@ struct Config {
         noTrie(false),
         noHopCache(false),
         writeStats(false),
-        gridSize(2000 / util::geo::M_PER_DEG) {}
+        gridSize(2000 / util::geo::M_PER_DEG),
+        gaussianNoise(0) {}
   std::string dbgOutputPath;
   std::string solveMethod;
   std::string shapeTripId;
@@ -60,6 +61,7 @@ struct Config {
   bool noHopCache;
   bool writeStats;
   double gridSize;
+  double gaussianNoise;
 
   std::string toString() {
     std::stringstream ss;

src/pfaedle/router/ShapeBuilder.cpp

@@ -167,19 +167,25 @@ EdgeCandGroup ShapeBuilder::getEdgCands(const Stop* s) const {
   auto pos = POINT(s->getLng(), s->getLat());
   ret.push_back({0, 0, 0, pos, 0, {}});
 
-  // unsigned seed =
-  // std::chrono::system_clock::now().time_since_epoch().count();
-  // std::default_random_engine gen(seed);
-  // std::normal_distribution<double> dist(0.0, 25.0);
-
-  // add some gaussian noise
-  // pos.setX(pos.getX() + dist(gen));
-  // pos.setY(pos.getY() + dist(gen));
-
   double maxMDist = _motCfg.osmBuildOpts.maxStationCandDistance;
 
   double distor = util::geo::latLngDistFactor(pos);
 
+  if (_cfg.gaussianNoise > 0) {
+    unsigned seed = std::chrono::system_clock::now().time_since_epoch().count();
+    std::default_random_engine gen(seed);
+
+    // the standard dev is given in meters, convert (roughly...) to degrees
+    double standardDev = (_cfg.gaussianNoise / M_PER_DEG) / distor;
+
+    // mean 0 (no movement), standard dev according to config
+    std::normal_distribution<double> dist(0.0, standardDev);
+
+    // add gaussian noise
+    pos.setX(pos.getX() + dist(gen));
+    pos.setY(pos.getY() + dist(gen));
+  }
+
   std::set<trgraph::Node*> frNIdx;
   _nGrid.get(util::geo::pad(util::geo::getBoundingBox(pos),
                             (maxMDist / M_PER_DEG) / distor),
@@ -594,7 +600,7 @@ ad::cppgtfs::gtfs::Shape ShapeBuilder::getGtfsShape(
 std::string ShapeBuilder::getFreeShapeId(Trip* trip) {
   std::string ret;
   std::lock_guard<std::mutex> guard(_shpMutex);
-  while (!ret.size() || _feed->getShapes().get(ret)) {
+  while (!ret.size() || _feed->getShapes().has(ret)) {
     _curShpCnt++;
     ret = "shp_";
     ret += std::to_string(trip->getRoute()->getType());

src/shapevl/Collector.cpp

@@ -169,7 +169,7 @@ std::vector<LINE> Collector::segmentize(const Trip* t, const LINE& shape,
   // 5) As tracks are often longer than 20 meters, this will dillute our AN
   //    measure, although the shape is CORRECT (because the ground truth uses
   //    a different position philosophy than the test data)
-  // 6) To normalize this, we always the following approach:
+  // 6) To normalize this, we use the following approach:
   //      a) Get the exact progression of the measurment on the shape
   //      b) Extract a segment of 200 meters, with the measurement progress in
   //      the middle c) Project the GROUND TRUTH station coordinate to this

src/shapevl/ShapevlMain.cpp

@@ -56,8 +56,11 @@ void eval(const std::vector<std::string>* paths,
     for (const auto& oldTrip : evalFeed->getTrips()) {
       if (!mots->count(oldTrip.second->getRoute()->getType())) continue;
       auto newTrip = feed.getTrips().get(oldTrip.first);
-      if (!newTrip)
-        LOG(ERROR) << "Trip #" << oldTrip.first << " not present in " << path;
+      if (!newTrip) {
+        LOG(ERROR) << "Trip #" << oldTrip.first << " not present in " << path
+                   << ", skipping...";
+        continue;
+      }
       (*colls)[myFeed].add(oldTrip.second, oldTrip.second->getShape(), newTrip,
                            newTrip->getShape());
     }
@@ -121,6 +124,7 @@ int main(int argc, char** argv) {
     evlFeedPaths.push_back(feedPath);
     if (fullReportPath.size()) {
       reportStreams.emplace_back();
+      reportStreams.back().exceptions(std::ios::failbit | std::ios::badbit);
       reportStreams.back().open(fullReportPath + "/" +
                                 util::split(feedPath, '/').back() +
                                 ".fullreport.tsv");