Deliberative and Conceptual Inference in Service Robots
Round 1
Reviewer 1 Report
The paper presents an approach to service robotics aiming at integrating deliberative and inference processes based on the SitLog framework. SitLog represents a "programming environment" based on Prolog used to integrate robot behaviors (or skills) and realize complex "social tasks".
The authors address a complex research problem touching a number of aspects ranging from knowledge representation perception, complex decision making and automated planning and execution of plans. Although interesting there is a number of issues the authors should consider to improve the quality of the work.
One of the main issue in my opinion is a lack of discussion of related works to situate the contribution with respect to the state of the art. A real motivation of the proposed approach with respect to existing technologies and approach is missing. In this sense, it is not easy to capture the actual novelty of the approach and thus the advancement with respect to the state of the art.
A more detailed discussion of related works in social robotics, knowledge representation and planning and also cognitive architectures and human-robot (or more in general human-machine) interaction literature is therefore necessary to better motivate and contextualize the work. At the end of the review there is a list of related works the authors should consider to improve the related works section.
For example, the extensive use of prolog as basis to the present SitLog programming framework seems a bit outdated with respect to the available technologies. A better discussion of the motivations of the proposed approach is strictly necessary. For example it is not clear the why the authors do not consider the use of automated planning technologies like e.g., PDDL-based planners to support decision making and the synthesis of plans. There are ready-to-use technologies that seem more reliable, efficient and also expressive (e.g., temporal planners support support concurrency and time constraints) with respect to prolog-based reasoning.
Similarly, concerning knowledge representation it is not clear why the authors do not use standard semantic technologies that may facilitate integration, reuse of existing dictionaries as well as ontologies that deal with common-sense knowledge and similar concepts. In this regard, the section dedicated to knowledge representation is not much clear and there is a bit of confusion about the actual contribution of the authors. It is not clear if the authors have defined a "simple" taxonomy of concepts (i.e., a hierarchical structure of concepts characterized by "is-a" relationships) or a more complex ontology specifying "classes" with general properties with for instance cardinality constraints between concepts. A better discussion of these aspects and a more detailed description of the actual knowledge defined and the underlying assumptions is necessary.
Another issue concerns the generality of the proposed approach which seems strictly tailored to the considered scenarios. It is not easy to extract general functioning mechanisms that can be replicated in different scenarios. For example, concerning the integration of knowledge representation and planning there are some works that show how the integration of these techniques can be use to enhance the flexibility or the autonomy level of robot behaviors. On the one hand knowledge reasoning can be used to support common sense knowledge and facilitate the planing process by grounding predicates according to objects that can be used to carry out some types of tasks [7]. On the other, knowledge reasoning can be used to abstract perception information and allow a robot to autonomously recognize complex situations that may trigger planning goals to support proactive behaviors [16]. Similarly, knowledge representation can support socially-compliant behaviors and thus support the planning mechanism in performing and executing contextualized tasks/behaviors, adapted to the social dynamics [14]. The authors should better emphasize general mechanisms concerning the proposed architecture that are relevant with respect to the synthesis of autonomous behaviors of robots in social contexts.
Finally, a more extensive experimental analysis would be necessary to better show the efficacy and the scalability of the approach. An evaluation of human feedback is completely messing. Although in simulation, an evaluation of the interaction and the effectiveness of the observed robot behaviors from the human-perspective is necessary.
==============================
[1] K. Rajan, A. Saffiotti "Towards a science of integrated AI and Robotics". AIJ. 2017.
[2] F. Ingrand, M. Ghallab "Deliberation for autonomous robots: A survey". AIJ 2017.
[3] M. Ghallab, D. Nau, P. Traverso "The actor's view of automated planning and acting: A position paper". AIJ. 2014.
[4] F. Py, K. Rajan, C. McGann "A systematic agent framework for situated autonomous systems". AAMAS. 2010.
[5] M. Cialdea Mayer, A. Orlandini, A. Umbrico "Planning and execution with flexible timelines: a formal account". Acta Informatica. 2016.
[6] Fox, M.; and Long, D. 2003. PDDL2.1: An Extension toPDDL for Expressing Temporal Planning Domains.J. Artif.Intell. Res. (JAIR)20: 61–124
[7] Awaad, I., Kraetzschmar, G.K., Hertzberg, J. "The Role of Functional Affordances in Socializing Robots". SORO: 2015.
[8] Berlin, M., Gray, J., Thomaz, A.L., Breazeal, C. "Perspective Taking: An OrganizingPrinciple for Learning in Human-Robot Interaction". AAAI 2006.
[9] Laird, J.E., Newell, A., Rosenbloom, P.S. "SOAR: An architecture for general intelligence". AIJ. 1987.
[10] Langley, P., Laird, J.E., Rogers, S. "Cognitive architectures: Research issues and challenges". Cognitive Systems Research. 2009.
[11] Mataric, M.J., Scassellati, B. "Socially Assistive Robotics". 2016.
[12] Rossi, S., Ferland, F., Tapus, A. "User profiling and behavioral adaptation for HRI: A survey". 2017.
[13] R. De Benedictis, A. Umbrico, F. Fracasso, G. Cortellessa, A. Orlandini and A. Cesta "A Two-Layered Approach to Adaptive Dialogues for Robotic Assistance”. IEEE RO-MAN 2020.
[14] B. Bruno, C. T. Recchiuto, I. Papadopoulos, A. Saffiotti, C. Koulouglioti,R. Menicatti, F. Mastrogiovanni, R. Zaccaria, and A. Sgorbissa, "Knowledge representation for culturally competent personal robots: Requirements, design principles, implementation, and assessment”. 2019.
[15] A. Lieto, M. Bhatt, A. Oltramari, and D. Vernon "The role of cognitive architectures in general artificial intelligence". 2018.
[16] A. Umbrico, A. Cesta, G. Cortellessa, and A. Orlandini "A holistic approach to behavior adaptation for socially assistive robots". SORO. 2020
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper presents a deliberative and conceptual inference in service robots. The authors have proposed two strategies to implement the common sense daily life inference cycle: a pipe-line strategy and the use of the knowledge. To check the both approaches, the authors make model and compare the strengths and limitations. Overall, the paper is well written and organized. In addition, the contribution of the paper is clear. Therefore, the paper is worth being accepted. Follows are minor comments:
- The size of characters in all figures are too small to see. This needs to be checked.
- There are some spell errors. Typos are need to be checked.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
