This paper provides a quick overview of a Knowledge Engineering system, called ROS-TiPlEx (Timeline-based Planning and Execution with ROS), to provide a shared environment in which experts in robotics and planning can easily interact to, respectively, encode information about low-level robot control and de ne task planning and execution models. ROS-TiPlEx aims at facilitating the interaction between both kind of experts, thus, enhancing and possibly speeding up the process of an integrated control design. ROS-TiPlEx is the rst tool addressing the connection of ROS and timeline-based planning.
Deploying interactive robots in human-populated scenarios requires to address
multiple challenges. Among others, it is of paramount importance the ability of
the robot to quickly adapt its behaviours to the actual state of the environment
and to keep the user engaged in the interaction. Such highly exible and
adaptive behaviours are necessary also to guarantee safe and e ective human-robot
interactions. There are several approaches that aim to achieve robust action
selection via planning, e.g., [
Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
In real-world application the design of a Human-Robot Interaction (HRI) system, presents various needs that can not be ful lled by a single person. The main issue is the variety of information that must be processed all together to reach the nal result. Often the contribute of the expert is speci c to its eld and not integrated with the rest of the solutions; each actor takes in charge single building-blocks of the process to which he contributes with his expertise. The main role in the construction of such scenario is focused on the mission expert, or rather the person that knows the mission objectives, and is able to dispatch the work to the engineers that best match the competencies for the speci c task. Such an expert will be responsible for the correct execution of the separated tasks and it will be the main point of contact in case of any issue in the project. But if the mission leader is responsible for the overall success of the project, then other gures will have responsibility on the single building-blocks of the system. For the sake of brevity, we can summarize the main gure of expertise in a planning expert and a robot expert. The rst has knowledge of P&S, and can provide robust planning for the system, to reach the standard of quality and safety that must be met for any HRI application; the second, on the other hand, is the one responsible for the robot itself. He should provide hardware capable of satisfying the mission objective and equipped with all the tool that is needed by the speci c application (e.g. robot arms, camera, working tools, etc...). If we want to summarize the main roles and tasks: 1. Mission leader : Responsible for the whole project and in charge of dispatching the tasks to the people with the matching competencies; 2. Planning expert : In charge of creating a robust P&S domain to reach the standard of quality and safety imposed by the mission leader; also responsible for the validation and real-time execution of the plan; 3. Robot expert : In charge of the hardware platform that must provide all the tool needed to accomplish the mission objectives de ned by the mission leader. It is also responsible for the quality of the hardware.
Such di erent roles will most likely have di erent background and di erent vision of the system. Not to mention other constraints like geographic position (the experts could be in di erent parts of the globe). If we consider the heterogeneity of such scenario, it is clear the need for a shared framework that helps the di erent actors to collaborate, share the same vision, and have a common platform where they can communicate. ROS-TiPlEx aims to ll the gap and promotes communication among these di erent gures, even though in its rst implementation it can`t support all the actors in the integrity of the process, even though automates the most part of it. 3
The context. ROS-TiPlEx([
The architecture. ROS-TiPlEx is designed with a Layered architecture, that allows introducing progressive levels of abstraction throughout the process, which lead to the modularity of the whole process. The process starts from Layer 0, that is in contact with the ROS platform (i.e., the robotic platform), and is able to access its data. In the speci city of ROS, layer 0 behaves as a publisher and subscriber that registers to the ROS topics involved in the execution of the mission tasks. Once the system is scanned, the process can pass to Layer 1, where the robot expert should start his activity. Knowing the robot and the premises given by the mission leader, he should rstly de ne an atomic action of the robot (e.g. move from A to B), and map such action on an FSM (Finite State Machine), where the edges correspond to the topics involved in the action, and the states are unique names decided by the expert in relation to the speci c domain or mission goals. In this FSM, when a message passes on the topic related to the edges, a change of state is triggered and re ected on the P&S system. The manipulations on this layer are allowed thanks to two components integrated into ROS-TiPlEx: the ROS-bridge component that allows the access to the system thanks to a Javascript UI, and the Mongo-DB component, that stores the data in the DB and allows them to be reused. These same components are the one used on Layer 2, granting now the access for a planning expert. He has access to the atomic actions created in the previous layer and can manipulate them, to transform the simple FSMs in more complex ones. Such conversion is still based on the speci cation given by the mission leader and is needed to prepare a suitable model for the P&S domain, that would allow improved performances and stability when it comes to elaborate the planning domain and the execution plan. After the rework performed by the planning expert, the information in the system goes through the last block, the Layer 3, whose purpose is to convert all the data in les that will serve as input for the P&S software. In the case of PLATINUm, the output will be composed by the domain description and some Java classes which still need to be updated, in order to allow the execution of the plan that will be dispatched by PLATINUm. 4
A simple yet signi cant use case has been deployed together with the
ROSTiPlEx, to test and evaluate the software. It is possible to run the software using
the code in [
In this rst implementation of ROS-TiPlEx, we have shown how it is possible to build an engineering tool that allows the communication and the information sharing of di erent entities, with di erent skills, such as a robotic expert and a P&S expert. Albeit this strategy has been presented and experimented, there still are improvements that will lead to a more complete and more usable tool. A rst advancement should be done on adding to ROS-TiPlEx a module that allows the mission leader to de ne an unambiguous description of the mission scenario so that the experts participating to the process can have a common vision of the nal result. Nonetheless, the shared point of information could be accessible from di erent geographical points, enabling the faculty to work on the same topic even without a physical presence. Another advancement needed in ROS-TiPlEx is to improve the UI implementation, that now is not very satisfactory when it comes to user experience or usability. Moreover, an overall refactoring of the code is needed to improve the separation of the front and the back end, enabling better portability and the faculty to split the components in multiple machines, centralizing on the robot only what is strictly needed to the elaboration and execution of the plan. Last but not least, ROS-TiPlEx could be equipped with one level more of abstraction, in order to make its implementation independent from the actual ROS version, because this OS is in constant evolution and a new tool can not be limited by the versioning of the underlying platform. In conclusion, the added value of ROS-TiPlEx is to bring a shared vision in the process of development related to the P&S applied to robotics, on the for of an engineering tool that is able to connect di erent experts with di erent needs, and help them to reach the common goal of the the mission scenario satisfaction.