Using Boosted Features for the Detection of People in 2D Range Data

Abstract

Abstract-This paper addresses the problem of detecting people in two dimensional range scans. Previous approaches have mostly used pre-defined features for the detection and tracking of people. We propose an approach that utilizes a supervised learning technique to create a classifier that facilitates the detection of people. In particular, our approach applies AdaBoost to train a strong classifier from simple features of groups of neighboring beams corresponding to legs in range data. Experimental results carried out with laser range data illustrate the robustness of our approach even in cluttered office environments. I. INTRODUCTIONDetecting people is a key capacity for robots that operate in populated environments. Knowledge about presence, position, and motion state of people will enable robots to better understand and anticipate intentions and actions.In this paper, we consider the problem of people detection from data acquired with laser range finders. The application of such sensors for this task has been popular in the past as they provide a large field of view and, opposed to vision, are mainly independent from ambient conditions. However, laser range data contain little information about people, especially because they typically consist of twodimensional range information. Figure 1 shows an example scan from a cluttered office environment. While this scan was recorded, several people walked through the office. The scan suggests that in cluttered environments, people detection in 2D is difficult even for humans. However, at a closer look, range measurements that correspond to humans have certain geometrical properties such as size, circularity, convexity or compactness (see Figure 2). The key idea of this work is to determine a set of meaningful scalar features that quantify these properties and to use supervised learning to create a people detector with the most informative features. In particular, our approach uses AdaBoost as a method for selecting the best features and thresholds, while at the same time creating a classifier using the selected features.In the past, many researchers focused on the problem of tracking people in range scans. One of the most popular approach in this context is to extract legs by the detecting moving blobs that appear as local minima in the range image [1], [2], [3], [4]. To this end, two types of features have been quite popular: motion and geometry features. Motion in range data is typically identified by subtracting two subsequent scans. If the robot is moving itself, the scans have first to be aligned, e.g., using scan matching. The drawback of motion features is that only moving people can be found. Topp and Christensen [5] extend the method of Schulz et

Low Resolution Lidar-Based Multi-Object Tracking for Driving Applications

“…Classic approaches for object detection in lidar point clouds use clustering algorithms to segment the data, assigning the resulting groups to different classes [2,27,6,18]. Other strategies, such as the one used as baseline method in this paper, benefit from prior knowledge of the environment structure to ease the object segmentation and clustering [20,24].…”

Tracking and intercepting pedestrians: a robotic approach to surveillance of critical infrastructures

“…Consequently, for each cluster, various geometric and statistical characteristics are extracted. The identification can be made either by manually selecting the fitting thresholds [21] or by using a machine learning approach [9].…”

“…The feature extraction can be considered as a function where the input is the set of points of the cluster in two-dimensional Cartesian coordinates and the output is a vector in n-dimensional space, f : R 2 → R n . Arras et al [9], proposes a scheme for indoor human detection from laser data that is considered a work of reference in the field [25,28]. They use a combination of geometrical, statistical and distance-dependent features.…”

“…A previously reported, detection algorithms based on the AdaBoost classifier use stumps as weak classifiers, and the response obtained is limited to a discrete outcome, namely two classes h t (s i ) : R n → {−1, +1} [9]. The implementation used in this work uses the Real AdaBoost version, which has been reported to be less immune to noise and to provide a better generalization than its discrete counterpart [35,36].…”

“…It has been used for pedestrian detection, by performing a segmentation of the laser points into clusters, and then using a set of features to classify whether each cluster belongs to a human or not [9]. Although this approach has been proven efficient indoors, it tends to be very scenespecific because it uses distance-based features.…”