Track-based self-supervised classification of dynamic obstacles

  title={Track-based self-supervised classification of dynamic obstacles},
  author={Roman Katz and Juan I. Nieto and Eduardo Mario Nebot and Bertrand Douillard},
  journal={Autonomous Robots},
This work introduces a self-supervised architecture for robust classification of moving obstacles in urban environments. Our approach presents a hierarchical scheme that relies on the stability of a subset of features given by a sensor to perform an initial robust classification based on unsupervised techniques. The obtained results are used as labels to train a set of supervised classifiers. The outcomes obtained with the second sensor can be used for higher level tasks such as segmentation or… 
A Self-Supervised Decision Fusion Framework for Building Detection
This study combines the previous supervised and unsupervised building detection frameworks to suggest a self-supervised learning architecture and borrows the major strength of the unsuper supervised approach to obtain one of the most important clues, the relation of a building, and its cast shadow.
Vision-based Autonomous Learning of Object Models on a Mobile Robot
This paper presents a novel approach that enables a mobile robot to autonomously learn object models using local, global and temporal visual cues and used the learned object models to reliably and efficiently recognize the desired objects in novel scenes.
Learning Object Models on a Robot using Visual Context and Appearance Cues
An algorithm that enables robots to use motion cues to identify (and focus on) a set of interesting objects, automatically extracting appearance-based and contextual cues from a small number of images to efficiently learn representative models of these objects.
Autonomous learning of vision-based layered object models on mobile robots
This paper presents an approach that enables a mobile robot to autonomously learn layered models for environmental objects using temporal, local and global visual cues.
Tracking and classification of dynamic obstacles with laser range finder in indoor environments
An approach to track and classify dynamic obstacles in indoor environments is proposed and a laser range finder (LRF) is used to detect and acquire information of dynamic obstacles.
Autonomous Learning of Object Models on Mobile Robots using Visual Cues
This dissertation describes an algorithm that investigates learning of object models from a small (3− 8) number of images to exploit complementary strengths of appearance-based and contextual visual cues to achieve the following objectives.
Moving Object Detection on the Side of Vehicle using Camera and LRF Sensor Fusion in Urban Environment
An advanced moving object detection algorithm in urban environment designed to reconstruct a three-dimensional world model and uses depth information and image data provided by laser scanners and cameras to calculate the ego-motion of the vehicle using internal sensors.
Unsupervised Change Detection in Heterogeneous Remote Sensing Imagery
The main novelty driving this study is that the comparison of affinity matrices can be used to define crossdomain similarities based on pixel relations rather than the direct comparison of radiometry values, which helps to tackle the problem of unsupervised heterogeneous change detection.
Dynamic Environment Perception for Mobile Robots
This thesis addresses the open question of learning probabilistic models that capture the dynamic aspects in spatial sensory data and investigates into the exploitation of these models towards a comprehension of non-stationarity.
Real-Time Fusion of Visual Images and Laser Data Images for Safe Navigation in Outdoor Environments
This research presents a probabilistic procedure for estimating the intensity and accuracy of the response of robots to earthquake-triggered disasters.


Classifying Dynamic Objects: An Unsupervised Learning Approach
An exemplar-based model for representing the time-varying appearance of objects in planar laser scans as well as a clustering procedure that builds a set of object classes from given training sequences are described.
Self-supervised Monocular Road Detection in Desert Terrain
This method for identifying drivable surfaces in difficult unpaved and offroad terrain conditions as encountered in the DARPA Grand Challenge robot race achieves robustness by combining sensor information from a laser range finder, a pose estimation system and a color camera.
Evaluating Feature Importance for Object Classification in Visual Surveillance
A method to evaluate the relative importance of various features for object type classification is described, which applies to the task of distinguishing whether an image blob is a vehicle, a single human, a human group, or a bike.
A Self-Supervised Terrain Roughness Estimator for Off-Road Autonomous Driving
A machine learning approach for estimating the second derivative of a drivable surface, its roughness, an evolution of one used in the 2005 DARPA Grand Challenge and analyzed using data from that route.
Probabilistic scheme for laser based motion detection
A stable, yet simple motion detection scheme that can be used and improved with tracking and classification procedures and a spatio-temporal correspondence procedure based on a scan registration algorithm is proposed.
Self-Supervised Classification for Planetary Rover Terrain Sensing
A self-supervised classification approach to learning the visual appearance of terrain classes which relies on vibration-based sensing of wheel-terrain interaction to identify these terrain classes is proposed.
Context-based vision system for place and object recognition
A low-dimensional global image representation is presented that provides relevant information for place recognition and categorization, and it is shown how such contextual information introduces strong priors that simplify object recognition.
Towards automatic discovery of object categories
A method to learn heterogeneous models of object classes for visual recognition that automatically identifies distinctive features in the training set and learns the set of model parameters using expectation maximization.
A Probabilistic Exemplar Approach to Combine Laser and Vision for Person Tracking
  • D. Schulz
  • Computer Science
    Robotics: Science and Systems
  • 2006
The experimental results show that the appearance of persons in camera image scan be tracked reliably using this approach and that it also allows to distinguish between persons during tracking.