Lifelong localization in changing environments

Abstract

Robot localization systems typically assume that the environment is static, ignoring the dynamics inherent in most real-world settings. Corresponding scenarios include households, offices, warehouses and parking lots, where the location of certain objects such as goods, furniture or cars can change over time. These changes typically lead to inconsistent observations with respect to previously learned maps and thus decrease the localization accuracy or even prevent the robot from globally localizing itself. In this paper we present a sound probabilistic approach to lifelong localization in changing environments using a combination of a Rao-Blackwellized particle filter with a hidden Markov model. By exploiting several properties of this model, we obtain a highly efficient map management approach for dynamic environments, which makes it feasible to run our algorithm online. Extensive experiments with a real robot in a dynamically changing environment demonstrate that our algorithm reliably adapts to changes in the environment and also outperforms the popular Monte-Carlo localization approach.

Modeling motion patterns of dynamic objects by IOHMM

“…For example, people normally walk straight in the corridor and only turn sideways in cases like entering the kitchen or an office. It is obviously valuable to be able to capture these regular tracks and utilize the information contained in the 'habitual' behavior in various potentially beneficiary robotic tasks such as long-term mapping and/or localization with dynamic objects as well as navigation in human populated environments involving avoiding or interacting with people such as [9]. …”

“…Although the HMM has been shown to be a powerful tool in dynamic occupancy grid mapping [8] [9], the transition parameters in conventional HMMs are fixed after training, which results in a homogeneous Markov chain. However, dynamic objects, such as pedestrians or vehicles, usually move in a manner that the motion pattern changes from time to time.…”

Life-long spatio-temporal exploration of dynamic environments

“…Recently, some authors proposed to exploit these conflicting measurements in order to obtain information about the world dynamics and proposed representations that model the environment dynamics explicitly. These dynamic representations have shown their potential by improving mobile robot localization in changing environments [3], [4], [5], [6].…”

“…Churchill and Newman [3] propose to cluster similar observations at the same spatial locations to form 'experiences' which are then associated with a given place and show that this approach improves autonomous vehicle localization. The authors of [5] represent the states of the environment components (cells of an occupancy grid) with a hidden Markov model and show that their representation improves the localization robustness as well. Kucner's method [20] learns conditional probabilities of neighbouring cells of an occupancy grid to model typical motion patterns in dynamic environments.…”

Long-term topological localisation for service robots in dynamic environments using spectral maps

“…The paper [17] presents a method on which the robot adapts its environment model every time it visits a place finding those features that are more stable and "forgetting" those that are less useful. [18] proposes a method in which a dynamic occupancy grid is used that distinguishes between highly dynamical objects, objects that can be moved around and objects that are static.…”