Abstract Over the last two decades, several deployments of robots for in-house assistance older adults have been trialled. However, these solutions are mostly prototypes and remain unused in real-life scenarios. In this work, we review historical current landscape field, to try understand why yet succeed as personal assistants daily life. Our analysis focuses on complementary aspects: capabilities physical platform logic deployment. The former shows regularities hardware configurations...

Pedestrian trajectory prediction is important in the research of mobile robot navigation environments with pedestrians. Most pedestrian algorithms require input historical trajectories to be complete. If a unobservable any frame past, then its become incomplete, algorithm will not predict future trajectory. To address this limitation, we propose STGN-IT, spatio-temporal graph network allowing incomplete input, which can pedestrians trajectories. STGN-IT uses an additional encoding method...

Designing a robotic application is challenging task. It requires vertical expertise spanning various fields, starting from hardware and low-level communication to high-level architectural solution for distributed applications. Today single expert cannot undertake the entire effort of creating robust reliable application. The current landscape robotics middlewares, ROS in primis, does not offer this problem yet; developers are expected be both designers domain experts. In our previous works...

In recent years, two fields have become more prominent in our everyday life: smart cities and service robots. a city, information is collected from distributed sensors around the city into centralised data hubs used to improve efficiency of systems provide better services citizens. Exploiting major advances Computer Vision Machine Learning, robots evolved performing simple tasks playing role hotel concierges, museum guides, waiters cafes restaurants, home assistants, automated delivery...

The focus of this work is to exploit ontologies make robotic systems more accessible non-expert users, therefore supporting the deployment robot-integrated applications. Due increasing number platforms available for commercial use, are nowadays being approached by users with different backgrounds, who often interested in robots' high-level capabilities than their technical architecture. Without right expertise however, using robots restricted exposed platform provider, i.e. they can only be...

To assist humans with their daily tasks, mobile robots are expected to navigate complex and dynamic environments, presenting unpredictable combinations of known unknown objects. Most state-of-the-art object recognition methods unsuitable for this scenario because they require that: (i) all target classes beforehand, (ii) a vast number training examples is provided each class. This evidence calls novel handle classes, which fewer images initially available (few-shot recognition). One way...

Modern robotic systems are a combination of sophisticated software and hardware components they offer complex functionalities. While popular middlewares that promote component-level reusability assist development already exist, there no established techniques or procedures use formal approach to robot system architecture design yet. This work aims at the long term goal model-based (and their architectures), by surpassing current based on personal expertise, best practices, in favor purely...

This paper describes a framework to extend the 3D robotic simulation environment Gazebo, and similar ones, with enhanced, tailor-made, multi-body dynamics specified in Modelica language.The body-to-body interaction models are written Modelica, but they use sophisticated collision detection capabilities of Gazebo engine.This contribution is first step toward complex robotics systems integrating detailed physics modelling realistic sensors such as lidar cameras.A proof-of-concept...

This paper describes the development of a Modelica simulator an All-Terrain vehicle using Vehicle Dynamics Library. The model aims at replicating behaviour Yamaha Grizzly 700 ATV, focusing on dynamics that are relevant to design and test control system. A thorough experimental validation simulator, demonstrating its accuracy in reproducing system, is also presented.

Despite the recent attitude toward making modern urban cities accessible, many barriers and obstacles still challenge impaired people mobility every day. Having a complete accurate map of both accessible paths could support drive planning renewing efforts. However, mapping timely, in an way, requires significant effort it is often left to volunteers no-profit associations. The Maps for Easy Paths (MEP) project represents attempt build accessibility maps (i.e., with routes) fusing, automatic...

Modelling complex systems is a common practice and de facto standard across most application domains in engineering. Although it would seem unreasonable - quite impractical to build structure as bridge without reference blueprint detailing how arrange all of its building blocks, Software Development, and, particularly the context Robotics, examples adhering rigorous modelling routines are still relatively rare find. Yet, models help understanding problems while pinpointing their potential...

This paper presents a novel approach to generating behavior trees for robots using lightweight large language models (LLMs) with maximum of 7 billion parameters. The study demonstrates that it is possible achieve satisfying results compact LLMs when fine-tuned on specific dataset. key contributions this research include the creation fine-tuning dataset based existing GPT-3.5 and comprehensive comparison multiple (namely llama2, llama-chat, code-llama) across nine distinct tasks. To be...

The results of scientific experiments performed by different groups are rarely directly comparable. Efforts such as the European Robotics League offer to community, in form competitions, well documented and stable benchmarks assess performance existing systems. However, can be equally useful at design time: Plug&Bench Benchmark Meta-model provides robot designers with a valuable addition their toolkit. Additionally, it enables -with benchmark composition-to predict system given individual components.

Service robots are expected to reliably make sense of complex, fast-changing environments. From a cognitive standpoint, they need the appropriate reasoning capabilities and background knowledge required exhibit human-like Visual Intelligence. In particular, our prior work has shown that ability reason about spatial relations between objects in world is key requirement for development Visually Intelligent Agents. this paper, we present framework commonsense which tailored real-world robotic...