Christoph Weinrich

Robot programming by demonstration through system identification

Increasingly, personalised robots - robots especially designed and programmed for an individuals needs and preferences - are being used to support humans in their daily lives, most notably in the area of service robotics. Arguably, the closer the robot is programmed to the individuals needs, the more useful it is, and we believe that giving people the opportunity to program their own robots, rather than programming robots for them, will push robotics research one step further in the personalised robotics field. However, traditional robot programming techniques require specialised technical skills from different disciplines and it is not reasonable to expect end-users to have these skills. In this paper, we therefore present a new method of obtaining robot control code - programming by demonstration through system identification - which algorithmically and automatically transfers human behaviours into robot control code, using transparent, analysable mathematical functions. Besides providing a simple means of generating perception-action mappings, they have the additional advantage that can also be used to form hypotheses and theoretical analysis of robot behaviour. We demonstrate the viability of this approach by teaching a Scitos G5 mobile robot to achieve wall following and corridor passing behaviours.

Estimation of human upper body orientation for mobile robotics using an SVM decision tree on monocular images

In this paper, we present a monocular, texture-based method for person detection and upper-body orientation classification. We build on a commonly used approach for person recognition that uses a Support Vector Machine (SVM) on Histograms of Oriented Gradients (HOG) [1] but replace the SVM by a decision tree with SVMs as binary decision makers. Thereby, in addition to the pure detection of persons, the distinction of eight upper-body orientation classes is enabled. The detection of humans and the estimation of their upper-body orientation from larger distances is essential for socially acceptable navigation of mobile robots. It permits to estimate the humans notice of the robot or even the humans interest in an interaction. Thus, it is the basis for the decision whether to approach or to avoid a human. By using an SVM decision tree for upper-body orientation estimation in discrete steps of 45°, we were able to classify about 64% of the test samples with an absolute error of less than 22.5°. This performance is much better than the results we obtained with comparable methods. Furthermore, our approach proved to be faster than the other state-of-the-art methods. This is of high relevance for implementation on mobile robots with limited computational resources.

ROREAS: robot coach for walking and orientation training in clinical post-stroke rehabilitation--prototype implementation and evaluation in field trials

This paper describes the objectives and the state of implementation of the ROREAS project which aims at developing a socially assistive robot coach for walking and orientation training of stroke patients in the clinical rehabilitation. The robot coach is to autonomously accompany the patients during their exercises practicing their mobility skills. This requires strongly user-centered, polite and attentive social navigation and interaction abilities that can motivate the patients to start, continue, and regularly repeat their self-training. The paper gives an overview of the training scenario and describes the constraints and requirements arising from the scenario and the operational environment. Moreover, it presents the mobile robot ROREAS and gives an overview of the robot’s system architecture and the required human- and situation-aware navigation and interaction skills. Finally, it describes our three-stage approach in conducting function and user tests in the clinical environment: pre-tests with technical staff, followed by function tests with clinical staff and user trials with volunteers from the group of stroke patients, and presents the results of these tests conducted so far.

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

People detection in 2D laser range data is a popular cue for person tracking in mobile robotics. Many approaches are designed to detect pairs of legs. These approaches perform well in many public environments. However, we are working on an assistance robot for stroke patients in a rehabilitation center, where most of the people need walking aids. These tools occlude or touch the legs of the patients. Thereby, approaches based on pure leg detection fail. The essential contribution of this paper are generic distance-invariant range scan features for people detection in 2D laser range data and the distinction of their walking aids. With these features we trained classifiers for detecting people without walking aids (or with crutches), people with walkers, and people in wheelchairs. Using this approach for people detection, we achieve an F1 score of 0.99 for people with and without walking aids, and 86% of detections are classified correctly regarding their walking aid. For comparison, using state-of-the-art features of Arras et al. on the same data results in an F1 score of 0.86 and 57% correct discrimination of walking aids. The proposed detection algorithm takes around 2.5% of the resources of a 2.8 GHz CPU core to process 270° laser range data at an update rate of 10 Hz.

People Tracking on a Mobile Companion Robot

Developing methods for people tracking on mobile robots is of great interest to engineers and scientists alike. Plenty of research is focused on pedestrian tracking in public areas. Yet, fewer work exists on practical people tracking in home environments with non-static cameras. This paper presents a real time people tracking system for mobile robots that filters asynchronous, multi-modal detections using a Kalman filter for each person. It allows for upright and sitting pose people tracking in home environments. We evaluate the performance of the tracking system using different detection modalities and compared it to state-of-the-art people detection methods. Evaluation was done on a newly collected indoor data set which we made publicly available for comparison and benchmarking.

Prediction of human collision avoidance behavior by lifelong learning for socially compliant robot navigation

In order to act socially compliant with humans, mobile robots need to show several behaviors that require the prediction of peoples motion. For example, when a robot avoids a person, it needs to respect the humans personal space [1] and the avoidance behavior needs to be smooth, so that it is understandable to the interaction partner. To achieve this, the robot needs to reason about future paths a person is likely to follow. Because humans adapt their avoidance behavior to the robots motion, the proposed method performs lifelong learning of the peoples behavior while it adapts its own behavior to their motion. The human avoidance behavior is modeled by a discrete, multi-modal, spatio-temporal distribution over the peoples future occurrences. This prediction is based on the peoples positions and their velocities relatively to the robot and the obstacle situation of the robots environment. The proposed prediction method is significantly better than a simple linear prediction. Particularly, for tactical decisions, like whether to avoid a moving person on the left or on the right side, this approach is well suited. Furthermore, when the humans get used to a robot, also a long-term change of the human behavior towards the robot can be learned by our approach.

Multi-modal people tracking on a mobile companion robot

People detection and tracking are key aspects in current research on mobile robots. While plenty of research is focused on pedestrian tracking in public areas, fewer work exists on practical people tracking in home environments with non static cameras. This paper presents a real-time people tracking system for mobile robots that applies multiple asynchronous detection modules and an efficient Kalman filter. It allows for upright pose- and under restrictions, sitting pose- people tracking in home environments. We evaluate the performance of the tracking system using different detection modalities on newly collected indoor data sets. These data sets are made publicly available for comparison and benchmarking.

Appearance-Based 3D Upper-Body Pose Estimation and Person Re-identification on Mobile Robots

In the field of human-robot interaction (HRI), detection, tracking and re-identification of humans in a robots surroundings are crucial tasks, e. g. for socially compliant robot navigation. Besides the 3D position detection, the estimation of a persons upper-body orientation based on monocular camera images is a challenging problem on a mobile platform. To obtain real-time position tracking as well as upper-body orientation estimations, the proposed system comprises discriminative detectors whose hypotheses are tracked by a Kalman filter-based multi-hypotheses tracker. For appearance-based person recognition, a generative approach, based on a 3D shape model, is used to refine these tracked hypotheses. This model evaluates edges and color-based discrimination from the background. Furthermore, for each person the texture of his or her upper-body is learned and used for person re-identification. When computational resources are limited, the update rate of the model-based optimization reduces itself automatically. Thereby the estimation accuracy decreases, but the system keeps tracking the persons around the robot in real-time. The persons 3D pose is tracked up to a distance of 5.0 meters with an average Euclidean error of 18 cm. The achieved motion independent average upper-body orientation error is 22°. Furthermore, the upper-body texture is learned on-line which allowed a stable person re-identification in our experiments.

Generic Distance-Invariant Features for Detecting People with Walking Aid in 2D Laser Range Data

People detection in 2D laser range data is a popular cue for person tracking in mobile robotics. Many approaches are designed to detect pairs of legs. These approaches perform well in many public environments. However, we are working on an assistance robot for stroke patients in a rehabilitation center, where most of the people need walking aids. These tools occlude or touch the legs of the patients. Thereby, approaches based on pure leg detection fail. The essential contribution of this paper are generic distance-invariant range scan features for people detection in 2D laser range data. The proposed approach was used to train classifiers for detecting people without walking aids, people with walkers, people in wheelchairs, and people with crutches. By the use of these features, the detection accuracy of people without walking aids increased from an \(F_1\) score of 0.85 to 0.96, compared to the state-of-the-art features of Arras et al. Moreover, people with walkers are detected with an \(F_1\) score of 0.95 and people in wheelchairs with an \(F_1\) score of 0.94. The proposed detection algorithm takes on average less then 1 % of the resources of a 2.8 GHz CPU core to process 270\(^{\circ }\) laser range data with an update rate of 12 Hz.