diff --git a/cartographer_ros/scripts/dev/publish_fake_random_landmarks.py b/cartographer_ros/scripts/dev/publish_fake_random_landmarks.py
new file mode 100755
index 0000000..919a748
--- /dev/null
+++ b/cartographer_ros/scripts/dev/publish_fake_random_landmarks.py
@@ -0,0 +1,191 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+
+# Copyright 2018 Magazino GmbH
+#                The Cartographer Authors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import itertools
+import random
+
+import rospy
+from tf import transformations
+from cartographer_ros_msgs.msg import LandmarkEntry, LandmarkList
+
+DESC = '''
+Samples a number of random landmarks at a specified rate.
+Can be used to test the timing of landmark input or the effect of erroneous
+landmarks with duplicate IDs (if --allow_duplicate_ids is set).
+
+For example:
+
+./publish_fake_random_landmarks.py  \\
+  --publish_period 0.1  \\
+  --id_vocabulary A B C  \\
+  --id_length 5  \\
+  --sample_period 1.0
+
+will publish empty landmark lists at 10Hz and random landmarks every second.
+IDs are also sampled, using the cartesian product of the provided "vocabulary".
+In the above example, a sampled ID could be e.g. "AACBC" (length=5).
+'''
+
+TOPIC = "landmark"
+
+
+class LandmarkSamplerOptions(object):
+
+  def __init__(self, *args, **kwargs):
+    self.allow_duplicate_ids = False
+    self.id_vocabulary = {}
+    self.id_length = 0
+    self.max_distance = 0.
+    self.num_landmarks = 0
+    self.rotation_weight = 0.
+    self.translation_weight = 0.
+    for name, arg in kwargs.items():
+      setattr(self, name, arg)
+
+
+class LandmarkIdSampler(object):
+
+  def __init__(self, id_vocabulary, id_length):
+    # Precompute all combinations of the symbols in the vocabulary.
+    # WARNING: can be huge, check before potentially blocking the system.
+    if len(id_vocabulary)**id_length > 1e6:
+      raise ValueError("ID sampling space is too large")
+    self.sampling_space = list(
+        itertools.product(*(id_vocabulary for i in range(id_length))))
+
+  def sample_id(self):
+    # Draw a random combination of symbols and stringify it.
+    random_index = random.randint(0, len(self.sampling_space) - 1)
+    sampled_id = "".join(self.sampling_space[random_index])
+    return sampled_id
+
+
+class LandmarkSampler(object):
+
+  def __init__(self, options):
+    if not isinstance(options, LandmarkSamplerOptions):
+      raise TypeError("expected LandmarkSamplerOptions")
+    self.options = options
+    rospy.loginfo("Initializing landmark ID sampler...")
+    self.landmark_id_sampler = LandmarkIdSampler(options.id_vocabulary,
+                                                 options.id_length)
+    self._sampled_ids = []
+
+  def random_landmark(self):
+    landmark = LandmarkEntry()
+    landmark.translation_weight = self.options.translation_weight
+    landmark.rotation_weight = self.options.rotation_weight
+    landmark.id = self.landmark_id_sampler.sample_id()
+    if landmark.id in self._sampled_ids:
+      if not self.options.allow_duplicate_ids:
+        rospy.logwarn("Ignoring duplicate ID: {}".format(landmark.id))
+        return None
+      else:
+        rospy.logwarn("Duplicate ID: {}".format(landmark.id))
+    self._sampled_ids.append(landmark.id)
+
+    vector = transformations.random_vector(3) * self.options.max_distance
+    landmark.tracking_from_landmark_transform.position.x = vector[0]
+    landmark.tracking_from_landmark_transform.position.y = vector[1]
+    landmark.tracking_from_landmark_transform.position.z = vector[2]
+
+    quaternion = transformations.random_quaternion()
+    landmark.tracking_from_landmark_transform.orientation.x = quaternion[0]
+    landmark.tracking_from_landmark_transform.orientation.y = quaternion[1]
+    landmark.tracking_from_landmark_transform.orientation.z = quaternion[2]
+    landmark.tracking_from_landmark_transform.orientation.w = quaternion[3]
+    return landmark
+
+  def random_landmark_list(self):
+    landmark_list = LandmarkList()
+    landmark_list.header.stamp = rospy.Time.now()
+    for _ in range(self.options.num_landmarks):
+      random_landmark = self.random_landmark()
+      if random_landmark is not None:
+        landmark_list.landmarks.append(random_landmark)
+    return landmark_list
+
+
+class SampledLandmarkPublisher(object):
+
+  def __init__(self, publish_period, sample_period, landmark_sampler_options):
+    self.landmark_sampler = LandmarkSampler(landmark_sampler_options)
+    rospy.loginfo("Publishing landmarks to topic: {}".format(TOPIC))
+    self.publisher = rospy.Publisher(TOPIC, LandmarkList, queue_size=1)
+    self.publish_timer = rospy.Timer(
+        rospy.Duration(publish_period), self.publish_empty_landmark_list)
+    self.sample_timer = rospy.Timer(
+        rospy.Duration(sample_period), self.publish_random_landmark_list)
+
+  def publish_random_landmark_list(self, timer_event):
+    self.publisher.publish(self.landmark_sampler.random_landmark_list())
+
+  def publish_empty_landmark_list(self, timer_event):
+    landmark_list = LandmarkList(rospy.Header(stamp=rospy.Time.now()), [])
+    self.publisher.publish(landmark_list)
+
+
+if __name__ == '__main__':
+  import argparse
+  parser = argparse.ArgumentParser(
+      description=DESC, formatter_class=argparse.RawTextHelpFormatter)
+  parser.add_argument("--translation_weight", type=float, default=1e5)
+  parser.add_argument("--rotation_weight", type=float, default=1e5)
+  parser.add_argument(
+      "--publish_period",
+      type=float,
+      default=0.1,
+      help="Baseline period for publishing empty landmark lists.")
+  parser.add_argument(
+      "--sample_period",
+      type=float,
+      default=5.,
+      help="Period at which randomly sampled landmarks are published.")
+  parser.add_argument(
+      "--num_landmarks",
+      type=int,
+      default=5,
+      help="The number of random landmarks published simultaneously.")
+  parser.add_argument(
+      "--max_distance",
+      type=float,
+      default=1.0,
+      help="Maximum distance of a random landmark to the tracking frame.")
+  parser.add_argument(
+      "--id_vocabulary",
+      nargs='+',
+      default={"a", "b", "c", "1", "2", "3"},
+      help="Set of symbols that can appear in random landmark IDs.")
+  parser.add_argument(
+      "--id_length",
+      type=int,
+      default=5,
+      help="The length of the random landmark IDs (number of symbols).")
+  parser.add_argument(
+      "--allow_duplicate_ids",
+      action="store_true",
+      help="Publish landmarks with IDs that have already been published.")
+
+  args, unknown = parser.parse_known_args(rospy.myargv()[1:])
+  rospy.init_node("landmark_sampler")
+
+  landmark_sampler_options = LandmarkSamplerOptions(**args.__dict__)
+  sampler = SampledLandmarkPublisher(args.publish_period, args.sample_period,
+                                     landmark_sampler_options)
+
+  rospy.spin()