In this work we present a user interface assisting non-expert users to design complex robot behaviours, hence facilitating the deployment of robot-integrated applications. Due to the increasing number of robotic platforms available for commercial use, robots are being approached by users with di erent backgrounds, whose interests lie in the high-level capabilities of the platforms rather than their technical architecture. Our interface allows non-experts to use a robot as a development platform, i.e. giving it high-level commands (e.g. autonomous navigation, vision, natural language generation) by relying on a basic ontology of high-level capabilities mapped on the robot low-level capabilities (e.g. communication, synchronisation, drivers) exposed by the most common robotic middleware ROS. To show how our work can sensibly reduce the e orts required for having robots achieving basic tasks, we propose a live demonstration in which the ISWC audience will remotely program a robot to achieve di erent tasks without any previous knowledge of ROS.
The goal of our work is to facilitate the usability of robotic platforms to nonexpert users, hence simplifying the time-consuming process of con guring and programming robotic platforms to integrate in real-world applications. In this demo, we present a user-friendly interface that we designed and implemented in order to allow users to make robots of di erent types and capabilities achieve basic tasks without having any previous knowledge of robot programming.
Our motivation stems from the growth of robotic platforms available for commercial use, which parallels to the increasing number of non-technical users that approach robots to exploit them as development platforms for their own purposes. For example, we are looking at integrating robots in a number of smart city applications, such as parking monitoring, building surveillance and garbage collection. Without the technical knowledge or the expertise of dedicated robot developers, we are limited to exploit the capabilities allowed by the commercial platforms (e.g. using a remote-controlled drone to record photos or videos), while more advanced usages supported by those systems could be exploited (e.g. programming the drone to autonomously survey an area).
These limitations could be overcome by integrating a middleware between the robots and an application, such as the ROS framework (Robot Operating System4) promoted by the robotics community with the aim to relieve developers from the management of low-level components. From an end-user perspective, ROS remains a low-level technical platform directly above a robot's hardware components. This means that programming a robot to achieve high-level tasks is still a time-consuming process, requiring advanced knowledge of robot-speci c ROS components, the ROS framework, and robotics in general.
Ontologies, on the other hand, have proven to be successful in facilitating use
of complex systems, e.g. sensor networks, smart products and smart homes [1{3].
Starting from the idea that an ontological representation of robot capabilities
could help abstracting from the low-level implementation of the robot, we
developed in [
The user interface, presented in Figure 1, is composed of three blocks: the Capability Panel (left block), the Construct Panel (right block) and the main Program Panel (middle block). All demo material, i.e. a video of the process, the ontology and code to perform the tasks, is available online5.
Capability Panel. This panel o ers the users the list of capabilities abstracted
from a robotic platform to which the system is connected. We refer the reader
to [
Construct Panel. The Construct Panel shows the user the set of programming constructs that he/she can choose to create the program to instruct the robot with. A program can be created as an imperative programming language, in which the atomic blocks are invocations of the available robot capabilities on the left panel (using the add button), and a set of conditional operators, such as if-then-else, while-do and the repeat statement. An additional no-operation statement can be employed to perform empty operations. The parameters of a capability can be used in the conditions, so to exploit any robot output to drive the program ow (e.g. moving forward until an object is detected). Program Panel. The Program Panel in the middle of Figure 1 shows the user the program he/she has created through constructs and capabilities of the respective panels. Once the user is satis ed with the set of commands, they are sent to the robot through a Dynamic module. If a read capability is requested, the Dynamic module relays to the interface the messages received from the robot. If a write capability is requested, the Dynamic module reads the parameters from the interface, structures them in a message and then sends them to the robot, which consequently performs the operations. 3
We plan a live demonstration where the audience will be able to remotely instruct a ground wheeled robot operating in an o ce environment (the Knowledge Media Institute) to solve a number of high-level tasks using the presented interface, i.e. creating a program allowing the robot to achieve a speci c task. We plan tasks to be of increasing di culty for the audience to get familiar with the interface, namely:
The goal of the demonstration will be to show users without previous robotics expertise that e orts and time required to program robots can be sensibly reduced through relying on ontological knowledge, hence making a step towards facilitating the development of applications in which robots are integrated. If time and conditions allow, we also plan to collect quantitative and qualitative data about the audience's performance (e.g. time taken to perform an action or robot precision) in view of a larger and thorough evaluation of our tool.
In the future, we will be o ering a set of ne-grained APIs for a more highlevel communication with the robot, following the structure of the user interface presented in the paper.