To Örebro University

State of the art mapping algorithms can produce high-quality maps. However, they are still vulnerable to clutter and outliers which can affect map quality and in consequence hinder the performance of a robot, and further map processing for semantic understanding of the environment. This paper presents ROSE, a method for building-level structure detection in robotic maps. ROSE exploits the fact that indoor environments usually contain walls and straight-line elements along a limited set of orientations. Therefore metric maps often have a set of dominant directions. ROSE extracts these directions and uses this information to segment the map into structure and clutter through filtering the map in the frequency domain (an approach substantially underutilised in the mapping applications). Removing the clutter in this way makes wall detection (e.g. using the Hough transform) more robust. Our experiments demonstrate that (1) the application of ROSE for decluttering can substantially improve structural feature retrieval (e.g., walls) in cluttered environments, (2) ROSE can successfully distinguish between clutter and structure in the map even with substantial amount of noise and (3) ROSE can numerically assess the amount of structure in the map.

In this extended abstract we present a novel dataset for benchmarking motion prediction algorithms. We describe our approach to data collection which generates diverse and accurate human motion in a controlled weakly-scripted setup. We also give insights for building a universal benchmark for motion prediction.

Dynamics is an inherent feature of reality. In spite of that, the domain of maps of dynamics has not received a lot of attention yet. In this book, we present solutions for building maps of dynamics and outline how to make use of them for motion planning. In this chapter, we present discuss related research question that as of yet remain to be answered, and derive possible future research directions. 

This paper describes 2D specifications of the IEEE Map Data Representation (MDR) standard and introduces the ongoing efforts for developing 3D MDR specifications. The 2D MDR standard is well-suited for current robot applications utilizing 2D maps, which may include autonomous road navigation, robotic logistic systems, defense and rescue robots, and service robots for personal or domestic applications. In addition to their original purpose as a common format for representing the spatial environment for robot navigation, the developed MDR standards are expected to contribute to promote development of good experimental methodologies for mobile robotics as a valuable tool for facilitating comparison and evaluation of maps obtained from different systems or methodologies.

Change and motion are inherent features of reality. The ability to recognise patterns governing changes has allowed humans to thrive in a dynamic reality. Similarly, dynamics awareness can also improve the performance of robots. Dynamics awareness is an umbrella term covering a broad spectrum of concepts. In this chapter, we present the key aspects of dynamics awareness. We introduce two motivating examples presenting the challenges for robots operating in a dynamic environment. We discuss the benefits of using spatial models of dynamics and analyse the challenges of building such models.

The task of building maps of dynamics is the key focus of this book, as well as how to use them for motion planning. In this chapter, we present a categorisation and overview of different types of maps of dynamics. Furthermore, we give an overview of approaches to motion planning in dynamic environments, with a focus on motion planning over maps of dynamics. 

The shared feature of the flow of discrete objects and continuous media is that they both can be represented as velocity vectors encapsulating direction and speed of motion. In this chapter, we present a method for modelling the flow of discrete objects and continuous media as continuous Gaussian mixture fields. The proposed model associates to each part of the environment a Gaussian mixture model describing the local motion patterns. We also present a learning method, designed to build the model from a set of sparse, noisy and incomplete observations. 

The key idea of modelling flow of discrete objects is to capture the way they move through the environment. One method to capture the flow is to observe changes in occupancy caused by the motion of discrete objects. In this chapter, we present a method to model and learn occupancy shifts caused by an object moving through the environment. The key idea is observe temporal changes changes in the occupancy of adjacent cells, and based on the temporal offset infer the direction of the occupancy flow.

Maps of dynamics can be beneficial for motion planning. Information about motion patterns in the environment can lead to finding flow-aware paths, allowing robots to align better to the expected motion: either of other agents in the environment or the flow of air or another medium. The key idea of flow-aware motion planning is to include adherence to the flow represented in the MoD into the motion planning algorithm’s sub-units (i.e. cost function, sampling mechanism), thereby biasing the motion planner into obeying local and implicit traffic rules. 

Models of human behaviour, such as pedestrian flows, are beneficial for safe and efficient operation of mobile robots. We present a new methodology for benchmarking of pedestrian flow models based on the afforded safety of robot navigation in human-populated environments. While previous evaluations of pedestrian flow models focused on their predictive capabilities, we assess their ability to support safe path planning and scheduling. Using real-world datasets gathered continuously over several weeks, we benchmark state-of-theart pedestrian flow models, including both time-averaged and time-sensitive models. In the evaluation, we use the learned models to plan robot trajectories and then observe the number of times when the robot gets too close to humans, using a predefined social distance threshold. The experiments show that while traditional evaluation criteria based on model fidelity differ only marginally, the introduced criteria vary significantly depending on the model used, providing a natural interpretation of the expected safety of the system. For the time-averaged flow models, the number of encounters increases linearly with the percentage operating time of the robot, as might be reasonably expected. By contrast, for the time-sensitive models, the number of encounters grows sublinearly with the percentage operating time, by planning to avoid congested areas and times.