To Örebro University

We propose a context-aware navigation framework designed to support the navigation of autonomous ground vehicles, including articulated ones. The proposed framework employs a behavior tree with novel nodes to manage the navigation tasks: planner and controller selections, path planning, path following, and recovery. It incorporates a weather detection system and configurable global path planning and controller strategy selectors implemented as behavior tree action nodes. These components are integrated into a sub-tree that supervises and manages available options and parameters for global planners and control strategies by evaluating map and real-time sensor data. The proposed approach offers three key benefits: overcoming the limitations of single planner strategies in challenging scenarios; ensuring efficient path planning by balancing between optimization and computational effort; and achieving smoother navigation by reducing path curvature and improving drivability. The performance of the proposed framework is analyzed empirically, and compared against state of the art navigation systems with single path planning strategies.

Automated planning is a branch of artificial intelligence aiming at finding a course of action that achieves specified goals, given a description of the initial state of a system and a model of possible actions. There are plenty of planning approaches working under different assumptions and with different features (e.g. classical, temporal, and numeric planning). When automated planning is used in practice, however, the set of required features is often initially unclear. The Unified Planning (UP) library addresses this issue by providing a featurerich Python API for modeling automated planning problems, which are solved seamlessly by planning engines that specify the set of features they support. Once a problem is modeled, UP can automatically find engines that can solve it, based on the features used in the model. This greatly reduces the commitment to specific planning approaches and bridges the gap between planning technology and its users.

Integrated AI systems can often be composed of a series of black-box components. We propose a planning based approach that takes a set of components, initially available models/data, and set of goal models and automatically devices a plan that represents an integrated AI system. This allows us to automatically adjust to new components and changing requirements. Experimental evaluation shows how our approach performs over a large number of instances covering sets of over 100 components.

Behavior Trees (BTs) are a formalism increasingly used to control the execution of robotic systems. The strength of BTs resides in their compact, hierarchical and transparent representation. However, when used in practical applications transparency is often hindered by the introduction of implicit run-time relations between nodes, e.g., because of data dependencies or hardware-related ordering constraints. Manually verifying the correctness of a BT with respect to these hidden relations is a tedious and error-prone task. This paper presents a modular planning-based approach for automatically testing BTs offline at design time, to identify possible executions that may violate given data and ordering constraints and to exhibit traces of these executions to help debugging. Our approach supports both basic and advanced BT node types, e.g., supporting parallel behaviors, and can be extended with other node types as needed. We evaluate our approach on BTs used in a commercially deployed robotics system and on a large set of randomly generated trees showing that our approach scales to realistic sizes of more than 3000 nodes. 

As research in smart homes and activity recognition is increasing, it is of ever increasing importance to have benchmarks systems and data upon which researchers can compare methods. While synthetic data can be useful for certain method developments, real data sets that are open and shared are equally as important. This paper presents the E-care@home system, its installation in a real home setting, and a series of data sets that were collected using the E-care@home system. Our first contribution, the E-care@home system, is a collection of software modules for data collection, labeling, and various reasoning tasks such as activity recognition, person counting, and configuration planning. It supports a heterogeneous set of sensors that can be extended easily and connects collected sensor data to higher-level Artificial Intelligence (AI) reasoning modules. Our second contribution is a series of open data sets which can be used to recognize activities of daily living. In addition to these data sets, we describe the technical infrastructure that we have developed to collect the data and the physical environment. Each data set is annotated with ground-truth information, making it relevant for researchers interested in benchmarking different algorithms for activity recognition.

The demand for smart home technologies that enable ageingin place is rising. Through activity recognition, users’ activities can be monitored. However, for dementia patients, activity recognition alone cannot address the challenges associated with changes in the user’s habits along the disease’s stage transitions. Extending activity recognition to habit recognition enables the capturing of patients’ habits and change sin habits in order to detect anomalies. This paper aims to introduce relevant features for habit recognition solutions, extracted from data, in order to enrich the representation of the user’s habits. This solution is personalisable to meet the specific needs of the patients and generalizable for use in different scenarios. In this way caregivers are better informed on the expected changes of the patient’s habits, which can help to mitigate further deterioration through early treatment and intervention.

In many cases, complex multidisciplinary research projects may show a lack of coordinated development and integration, and a big effort is often required in the final phase of the projects in order to merge software developed by heterogeneous research groups. This is particularly true in advanced robotic projects: the objective here is to deliver a system that integrates all the hardware and software components, is capable of autonomous behaviour, and needs to be deployed in real-world scenarios toward providing an impact on future research and, ultimately, on society. On the other hand, in recent years there has been a growing interest for techniques related to software integration, but these have been mostly applied to the IT commercial domain.

This paper presents the work performed in the context of the project CARESSES, a multidisciplinary research project focusing on socially assistive robotics that involves 9 partners from the EU and Japan. Given the complexity of the project, a huge importance has been placed on software integration, task planning and architecture definition since the first stages of the work: to this aim, some of the practices commonly used in the commercial domain for software integration, such as merging software from the early stage, have been applied. As a case study, the document describes the steps which have been followed in the first year of the project discussing strengths and weaknesses of this approach.

The way in which humans behave, speak andinteract is deeply influenced by their culture. For example,greeting is done differently in France, in Sweden or in Japan;and the average interpersonal distance changes from onecultural group to the other. In order to successfully coexistwith humans, robots should also adapt their behavior to theculture, customs and manners of the persons they interact with.In this paper, we deal with an important ingredient of culturaladaptation: how to generate robot plans that respect givencultural preferences, and how to execute them in a way thatis sensitive to those preferences. We present initial results inthis direction in the context of the CARESSES project, a jointEU-Japan effort to build culturally competent assistive robots.

In this position paper we argue for moving towards generalpurpose domains to promote the usage of task planning forreal-world robot systems. Planning approaches should extractconcrete domains based on their current context in order tosolve problems. Towards this aim, we define the problem ofdomain reasoning, by which a planning domain is obtainedfrom a more general, multi-purpose domain definition, giventhe current deployment and context of the robot system. Weprovide examples motivating the need for domain reasoningin robot task planning, as well as a discussion of potentialsolutions to the domain reasoning problem.

This paper describes the design and implementationof a network scheduler prototype for IoT networks within the e-healthcare domain. The network scheduler combines a constraint-based task planner with the Lightweight Machine-to-Machine (LwM2M) protocol to be able to reconfigure IoT networks at run-time based on recognized activities and changes in the environment. To support such network scheduling, we implement a LwM2M application layer for the IoT devices that provides sensor data, network stack information, and a set of controllable parameters that affect the communication performance and the energy consumption.