People detection and distinction of their walking aids in 2D laser range data based on generic distance-invariant features

Weinrich, Christoph; Wengefeld, Tim; Schroeter, Christof; Groß, Horst–Michael

doi:10.1109/roman.2014.6926346

Cited by 38 publications

(39 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another solution is similar to the approach widely used in computer vision by extracting hand-crafted features and training classifiers. Image-based object detection is very popular in current autonomous driving research, such as road obstacles detection [6], mobility aids [7] and vehicle detection [8]. However, the sparse point data from a 2D Li-DAR are usually insufficient for reliable object identification using this kind of method within a single scan.…”

Section: Introductionmentioning

confidence: 99%

An Efficient L-Shape Fitting Method for Vehicle Pose Detection with 2D LiDAR

Chen

et al. 2018

2018 IEEE International Conference on Robotics and Biomimetics (ROBIO)

View full text Add to dashboard Cite

Detecting vehicles with strong robustness and high efficiency has become one of the key capabilities of fully autonomous driving cars. This topic has already been widely studied by GPU-accelerated deep learning approaches using image sensors and 3D LiDAR, however, few studies seek to address it with a horizontally mounted 2D laser scanner. 2D laser scanner is equipped on almost every autonomous vehicle for its superiorities in the field of view, lighting invariance, high accuracy and relatively low price. In this paper, we propose a highly efficient search-based L-Shape fitting algorithm for detecting positions and orientations of vehicles with a 2D laser scanner. Differing from the approach to formulating L-Shape fitting as a complex optimization problem, our method decomposes the L-Shape fitting into two steps: L-Shape vertexes searching and L-Shape corner localization. Our approach is computationally efficient due to its minimized complexity. In on-road experiments, our approach is capable of adapting to various circumstances with high efficiency and robustness.

show abstract

Section: Introductionmentioning

confidence: 99%

An Efficient L-Shape Fitting Method for Vehicle Pose Detection with 2D LiDAR

Chen

et al. 2018

2018 IEEE International Conference on Robotics and Biomimetics (ROBIO)

View full text Add to dashboard Cite

show abstract

“…Although Weinrich et al [34] provide their datasets, their recordings are limited to scenes with a single wheelchair or walker. While this might suffice for learning the few parameters of fairly constrained features and classifiers, we want to learn features from scratch and thus require more general and varied data.…”

Section: A Datasetmentioning

confidence: 99%

“…We trained our model on our training set and computed hyperparameters on our validation set as described in the previous section. In order to evaluate real-world performance of DROW, we now look at the precision and recall curves on our own test set and on the publicly available Reha test set of [34] 4 recorded with a similar robot. Recall that our test set was recorded in a never before seen part of the care facility, thus a method which learns location priors or background models will fail.…”

Section: Approach Evaluationmentioning

confidence: 99%

“…While early detection methods used simple heuristics such as fitting lines and circles [35], in the past few years hand crafted features, coupled with learned classifiers, have dominated laser based detection. Within this paradigm, a range of successful people [1], [30], [20], mobility aid [34] and road obstacle [18] detectors have been developed to support mobile robot navigation [8] and autonomous driving [28]. Even though the aforementioned models obtain respectable results, the general consensus seems to be that the information provided by 2D range data is not sufficient to reliably perform detection in a single scan, leading to approaches relying on sensor fusion [30], multilayered sensor setups [20], [29], or temporal integration of information by tracking.…”

Section: Introductionmentioning

confidence: 99%

“…Weinrich et al [34] face a similar task and propose the Gandalf detector for detecting people, wheelchairs, and walkers in individual laser scans. They introduce a new feature set for laser segments and classify these with an AdaBoost classifier.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

DROW: Real-Time Deep Learning-Based Wheelchair Detection in 2-D Range Data

Beyer

Hermans

Leibe

2017

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

We introduce the DROW detector, a deep learning based object detector operating on 2D range data. Laser scanners are lighting invariant, provide accurate 2D range data, and typically cover a large field of view, making them interesting sensors for robotics applications. So far, research on detection in laser 2D range data has been dominated by hand-crafted features and boosted classifiers, potentially losing performance due to suboptimal design choices. We propose a Convolutional Neural Network (CNN) based detector for this task. We show how to effectively apply CNNs for detection in 2D range data, and propose a depth preprocessing step and a voting scheme that significantly improve CNN performance. We demonstrate our approach on wheelchairs and walkers, obtaining state of the art detection results. Apart from the training data, none of our design choices limits the detector to these two classes, though. We provide a ROS node for our detector and release our dataset containing 464k laser scans, out of which 24k were annotated.

show abstract