**3. Results**

#### *3.1. Definitions and Types*

As part of robot categorization and definitions, the whole universe of robots can be divided into conventional and advanced robotics. Conventional robotics is also divided according to the fixation of the base; thus, one can identify mobile robots and robot manipulators, and conventional robotics also includes industrial robotics [2]. Following this idea, service robots belong to the field of advanced robotics. This general definition is valid but very broad, mainly because the field of robotics is a very vast research area, and proposing a general and universal definition is not a trivial task [3].

Towards the search of a better standardization, some committees have worked to develop an ontology that unifies entities in the field [4]. The ISO (International Organization for Standardization) organized an expert committee in 2007 to accomplish this task and the efforts materialized into the ISO 8373:2012 standard that defines a service robot as a robot that performs useful tasks for humans or equipment excluding industrial automation applications [5]. The International Federation of Robotics (IFR) agrees with such definition and appends that the robot needs to have a semi or fully autonomous behavior [6]. Other authors consider the incorporation of a natural communication system and the presence of artificial intelligence (AI) as a must in the category of service robots [7,8] to allow such robots to adapt to an uncontrollable environment. Moreover, the IEEE RAS (Robotics and Automation Society) has dedicated efforts in the task of unifying terms using advanced applied ontological methodologies [9] that aim to validate existing ontologies for consistency. By the time this work is written (2021), some advances exist on the generalization of definitions, entities, classes, and agents [10,11], but there is still work to be undertaken in this sense. Some benefits of standardization and generalization are the facilitation of design, production, knowledge, and technology transfer processes between groups and among the Research and Development areas to properly determine a paradigm in terminology and formalism.

The ISO standard, as well as the IFR [12], classifies service robots in two main classes, as can be observed in Figure 2. The first class mainly includes robots for personal and domestic use, such as robots that perform domestic tasks, entertainment robots, elderly and handicap assistance, personal transportation, home security and surveillance, and other types of domestic robots. The typical applications of these robots include tasks of non-commercial nature.

**Figure 2.** Taxonomy of service robots as proposed by the ISO 8373:2012.

Continuing with the classification, the second class is dedicated to service robots intended for professional use. The main taxonomy threads are field robotics, professional cleaning, inspections and maintenance systems, constructions and demolition, logistic systems, medical robotics, rescue and security applications, defense applications, underwater systems, and other professional service robots not specified above. The typical user for this kind of robot is an operator that has relevent former education or training in manual work. Contrary to personal robots, this specification involves commercial activities. The above classification can be better understood in Figure 3.

**Figure 3.** Robot categorization.

A complementary and more human-centric taxonomy is proposed by [9,13,14], where the main robotics definitions fall intro three classes depending on the human–robot interaction (HRI) level:


From the previous definitions arise some categorical problems. For instance, as pointed out by [9], an exoskeleton can be a medical robot when used in rehabilitation but can be a non-medical robot when used for assistance tasks, and can also be used in military tasks, contradicting the intended use of the robot. Further work is required to avoid ambiguous definitions, and ontology engineering provides the required framework for this disambiguation, as proposed by [15,16].

The technological stack of a service robot must address an uncontrolled environment and choose a clever combination of sensors in an integration effort to fulfill the required robot task to ensure a deep integration with humans [2]. Three main technical groups enable a service robot. These are software layers, contextualization, and human–robot interfaces. Software layers are primarily responsible for integrating the robotic device, connecting, and establishing a standard communication system for every component. Artificial intelligence is an important component of these robotic systems; tools such as TensorFlow or PyTorch execute machine learning algorithms and related tasks. Robotic systems need to know where they are and react properly based on their location, that is, to be aware of the spatial context of the environment [17]. This function is achieved via localization and sensorization. Finally, some human–robot interfaces implement its integration to the human workflow generally. Web dashboards, speech recognition, touchscreens, or even mobile device applications enable these interfaces. This technological stack is presented in a graphical form in Figure 4.

**Figure 4.** Common components of a service robot with examples of the technological stack available as in 2021.
