Sven Olufs

An efficient area-based observation model for monte-carlo robot Localization

The problem of mobile robot self-localization is considered as solved since Thruns et. al [1] pioneering work using monte-carlo filters for robot Localization (MCL). However, MCL is robust and precise under constraints like completely known environments and the sensor data must contain enough “true data” as contained in the map. In fact these conditions cannot always be guaranteed, which may results in a poor accuracy of the localization. In this paper we present a area-based observation model that is applied to MCL self-localization. The model is based on the idea of tracking the ground area inside the “free space” (not occupied cells) of a known map. Experimental data shows that the proposed model improves the robustness and accuracy of laser and stereo vision sensors under certain conditions like incomplete map, limited FOV and limited range of sensing. We also present an efficient approximation of our sensor model based on integral images.

Robust single view room structure segmentation in Manhattan-like environments from stereo vision

In this paper we propose a novel approach for the robust segmentation of room structure using Manhattan world assumption i.e. the frequently observed dominance of three mutually orthogonal vanishing directions in man-made environments. First, separate histograms are generated for the Cartesian major axis, i.e. X, Y and Z, on stereo data with an arbitrary roll, pitch and yaw rotation. Using the traditional Markov particle filters and minimal entropy as metric on the histograms, we are able to estimate the camera orientation with respect to orthogonal structure. Once the orientation is estimated we extract a hypotheses of the room structure by exploiting 2D histograms using mean shift clustering techniques as rough estimate for a pre-segmentation of voxels i.e. plane orientation and position. We apply superpixel over segmentation on the colour input to achieve a dense segmentation. The over segmentation and pre-segmented voxels are combined using graph-cuts for a not a-priori known number of final plane segments with a a-expansion graph cut variant proposed by Delong et al. with polynomial runtime. We show the robustness of our approach with respect to noise in real world data.

A simple inexpensive interface for robots using the Nintendo Wii controller

To have a robot at home might be great fun: it could fetch and carry things. However it remains open how to teach the robot the places it should go to in a manner that is cheap and entertaining for the user. This paper presents an easy-to-use interface that takes the robot on a virtual leash: using the Nintendo Wii remote the user can go towards target places while pointing at the robot. Using the inbuilt infrared camera and accelerometers and a couple of LEDs on the robot, the robot will follow the user. We show how a particle filter and an interacting multiple model (IMM) Kalman can be configured such that simple hand gestures with the Wii make the robot follow the users intention. The concept has been implemented on a mobile robot developed within the robotshome project. The robot leash interface has been tested with 12 volunteers who are interested in new technology but have never controlled a robot. The result is that most users could within a few minutes show the robot the first three places in a home environment. Given the little cost of the interface (about

Mobile Robot Obstacle Avoidance Based on Quasi-Holonomic Smooth Paths

50) the proposed robot leash is a promising human robot interface.

Embedded Stereo Vision

The paper explores the benefits of using continuous curvature 2D paths in obstacle avoidance methods in terms of safety and comfort. To this end the paper proposes a novel contribution by introducing clothoid-based smooth paths for non-holonomic robots defined from Cartesian (x,y) and angular velocities, as if they were intended to be applied to holonomic robots. To remark, these paths, coined as quasi-holonomic continuous curvature paths (QHCC), generate purely non-holonomic motions (combinations of linear and angular velocities) that mimic a holonomic motion. In fact, QHCC paths converge to the same asymptotic direction pointed by the holonomic motion while taking into account kinematic and dynamic constraints. In the paper, we show how these paths can be used and integrated in well-known obstacle avoidance algorithms such as Nearness Diagram (ND) and Dynamic Window Approach (DWA), among others. In addition to this, an ANN has been trained to considerably speed-up the path generation process and to learn the intrinsics of the path. Additionally, the paper shows the advantages of the proposed method over standard obstacle avoidance algorithms.

Object Detection and Classification for Domestic Robots

In the last few years, stereo vision has become a very interesting sensing technology for robotic platforms. Especially for indoor applications, stereo vision brings many advantages. It is passive, so it does not affect its environment, it is small in size, and it is very flexible. In this chapter, the possibilities of integrating stereo vision into smart cameras are explained. Stereo vision may involve a high algorithmic and computational effort, therefore not all stereo vision algorithms are suitable for real-time applications. This chapter gives a detailed description of two real-time stereo vision algorithms and presents possibilities to implement these concepts in a smart camera. Furthermore, it presents stereo vision using biologically inspired vision sensors for the depth estimation of scene dynamics.

Embedded vision-based Monte-Carlo robot localisation without additional sensors

A main task for domestic robots is to navigate safely at home, find places and detect objects. We set out to exploit the knowledge available to the robot to constrain the task of understanding the structure of its environment, i.e., ground for safe motion and walls for localisation, to simplify object detection and classification. We start from exploiting the known geometry and kinematics of the robot to obtain ground point disparities. This considerably improves robustness in combination with a histogram approach over patches in the disparity image. We then show that stereo data can be used for localisation and eventually for object detection classification and that this system approach improves object detection and classification rates considerably.

Room-structure estimation in Manhattan-like environments from dense 2½D range data using minumum entropy and histograms

This paper presents a fast approach for vision-based self-localisation in the RoboCup middle size league without additional e.g. dead reckoning sensors. An omni-directional vision system extracts a few features from image that are mapped to an sparse a-priori known map of the environment using Monte Carlo filters. The Monte Carlo filters are also used to model a virtual odometry (mass-inertia model) which is maintained through the filter itself. The precision of approach is directly compared to a traditional approach using the identical data. We show that the approach is stable and reactive while keeping the processing time low.

Robot on the Leash — An Intuitive Inexpensive Interface for Robots Using the Nintendo Wii Remote

In this paper we propose a novel approach for the robust estimation of room structure using Manhattan world assumption i.e. the frequently observed dominance of three mutually orthogonal vanishing directions in man-made environments. First, separate histograms are generated for every major axis, i.e. X, Y and Z, on stereo data with an arbitrary roll, pitch and yaw rotation. These histograms are maintained in the fashion of quadtrees. Using the traditional Markov particle filters and minimal entropy as metric on the histograms, we are able to estimate the camera orientation with respect to orthogonal structure. Once the orientation is estimated we extract hypothesis of the room structure by exploiting 2D histograms, i.e. X/Y, Z/Y, Z/X, using mean shift clustering techniques. Finally, the hypotheses are evaluated with the real data and false hypothesis are pruned. We also show the robustness of our approach with respect to noise in real world data.

Roboternavigation in Büros und Wohnungen

This paper presents an easy-to-use interface that takes the robot on a virtual leash: using the Nintendo Wii Remote the user can go towards target places while pointing at the robot. Using the inbuilt infrared camera and accelerometers and a couple of LEDs on the robot, the robot will follow the user. We show how a Particle Filter and an IMMKalman can be configured such that simple hand gestures with the Wii make the robot follow the user’s intention. The robot leash interface has been tested with 12 volunteers who are interested in new technology but have never controlled a robot. The result is that most users could within a few minutes show the robot the first three places in a home environment. Given the little cost of the interface (about