Jens Kessler

A graph matching technique for an appearance-based, visual SLAM-approach using Rao-Blackwellized Particle Filters

In continuation of our previous work on visual, appearance-based localization in manually built maps in this paper we present a novel appearance-based, visual SLAM approach. The essential contribution of this work is, an adaptive sensor model which is estimated online and a graph matching scheme to evaluate the likelihood of a given topological map. Both methods enable the combination of an appearance-based, visual localization concept with a Rao-Blackwellized Particle Filter (RBPF) as state estimator to a real-world suitable, online SLAM approach. In our system, each RBPF particle incrementally constructs its own graph-based environment model which is labeled with visual appearance features (extracted from panoramic 360deg snapshots of the environment) and the estimated poses of the places where the snapshots were captured. The essential advantages of this appearance-based SLAM approach are its low memory and computing-time requirements. Therefore, the algorithm is able to perform in real-time. Finally, we present the results of SLAM experiments in two challenging environments that investigate the stability and localization accuracy of this SLAM technique.

“Go Ahead, Please”: Recognition and Resolution of Conflict Situations in Narrow Passages for Polite Mobile Robot Navigation

For a mobile assistive robot operating in a human-populated environment, a polite navigation is an important requirement for the social acceptance. When operating in a confined environment, narrow passages can lead to deadlock situations with persons. In our approach we distinguish two types of deadlock situations at narrow passages, in which the robot lets the conflicting person pass, and either waits in a non-disturbing waiting position, or forms a queue with that person. Forthcoming deadlock situations are captured by a set of qualitative features. As part of these features, we detect narrow passages with a raycasting approach and predict the movement of persons. In contrast to numerical features, the qualitative description forms a more compact human-understandable space allowing to employ a rule-based decision tree to classify the considered situation types. To determine a non-disturbing waiting position, a multi-criteria optimization approach is used together with the Particle Swarm Optimization as solver. In field tests, we evaluated our approach for deadlock recognition in a hospital environment with narrow corridors.

I'll keep you in sight: Finding a good position to observe a person

Usually, in mobile robotics the robot has to deal with tasks like interacting with a person or performing a driving task. But what happens, if the robot just has to wait and thereby still has to react on user commands? In this case, the robot has to find a good position where the user can still be observed, and the robot does not disturb the users activities. Such a position has to fulfill different criteria: first, to guarantee the observability of the person with the robots on-board sensors and second, the robot should be able to observe the person when the person changes its position at its resting place. In this paper, a new approach is presented how to find a position, providing all these aspects, by solving an optimization problem using a particle swarm optimizer. We also present first results for that problem in the 2D and 3D case.

I'm still watching you: Update on observing a person in a home environment

Recently, a mobile robot was enabled to observe persons within their home in an unobtrusive way. This behavior is necessary during the long idle periods of the robot, where the robot has to stay somewhere near the user to be able to recognize new commands and tasks. Particle swarm optimization is used to find an optimal position. In this work, the ongoing progress of our approach is presented with focus on long term, real world capable improvements. Therefore, the online update of a person distribution and an elevation map are introduced. Finally, experiments show the robustness of the found optimal observation position.

Using a Spatio-Temporal FastMarching Planner to Politely Avoid Moving Persons

When mobile robots operate in home environments, a robot should consider the inhabitants while moving around. In this work, an approach is presented, which at the one hand predicts the movements of a person in a very simple way, and on the other hand uses the predicted movement to plan a motion path of the robot. We deploy a potential field approach to predict the person’s movement trajectory and use an modified Fast Marching planner to access a time-variable cost function for the planning process. The goal of our development is an early avoiding behavior of the robot, when the robot passes a person. This should increase the acceptance of the robot, and signal a “busy”-behavior. We show the feasibility of the presented approach in some first simulation results.