In the rapidly advancing landscape of robotics and artificial intelligence, we have the privilege of diving into the insights of Richard Duro, the project coordinator of Pillar-Robots from the University of A Coruña. Pillar-Robots stands as a beacon of innovation, aiming to create a future where autonomous robots seamlessly align with human desires in a safe and trustworthy manner, particularly in the domain of Mobile Manipulation.
What does Pillar-Robots envision for the future of autonomous robots, and how do you see this technology evolving, also in the sphere of Mobile Manipulation?
PILLAR-Robots is based on a vision of robots that autonomously learn to comply with the desires of humans in a safe and trustworthy manner. From this perspective, it addresses two key problems: the lifelong open-ended learning problem and the alignment problem. The first problem deals with the mechanisms for a robot to be able to operate and learn in domains that were not contemplated at design time, and to efficiently reuse previously learned knowledge in new domains. The second one addresses the problem of providing the robot with a purpose, that is, a general objective that is aligned with the users’ desires naturally. A purposeful robot will be able to glean from its experience and general indications from its users what would be worth achieving, and therefore, what it should seek to learn. This obviously leads to robots that do not need to be programmed for each task and that interact with humans naturally, even predicting future desires. Of course, the approach can be applied to any domain, including that of mobile manipulation. Instead of programming or teaching robots to perform specific manipulations, this approach would have the robot biasing its learning processes within any environment to favour learning what would be useful to achieve the manipulation objective of the user.
Looking ahead, what real-life applications do you anticipate for Pillar-Robots, and how might these applications transform industries?
Purpose-enabled lifelong open-ended learning robots will transform the way we view robots. They will not be just tools that we program to carry out specific tasks. They will become machines to enable the achievement of our desires, much like colleagues and friends. A Pillar-enabled robot will seek to achieve the purpose we set for it, whatever domain it finds itself in, and learning along the way whatever it may require satisfying our purpose for it. This opens whole new markets for robots in social domains, assistance, education, and others as well as a new way to define what a robot should do, independently of changes in the domain, within industrial settings. It is important to note here that this is especially relevant in the case of SMEs, which require robots with much higher capabilities of adapting to different tasks and domains, as they usually work with less structured fabrication facilities and much more diverse and customized products. Easily adaptable robots will reduce the cost and effort to robotize production and services.
How do you foresee Pillar-Robots enabling robots to be more creative in the future, and what possibilities might this create for industries and users?
Creativity is related to the ability to discover new and original ideas, connections, and solutions to problems. In general, it provides human beings with the capability to explore alternative approaches to problem-solving. In a certain sense, it results from a less constrained way of dealing with possible situations the system may find itself in. Thus, in this sense, any approach to Lifelong Open-ended Learning Autonomy (LOLA) fosters creativity from the viewpoint that the system will be searching for solutions to problems it has never seen before or problems that, even though it might have addressed in previous interactions, it will not have done so in the same domain or circumstances. This search is guided by a very general purpose, but the robot will have to autonomously define its particular needs and goals to satisfy that purpose in an extremely unconstrained fashion. It is expected that using this approach, many creative solutions will arise.
As Pillar-Robots progresses, how do you expect the relationship between humans and robots to develop, and what role will human input play in shaping the goals of these autonomous robots?
This is a very important question within the PILLAR-Robots project. In fact, it is one of the main issues we will be addressing: That of the alignment of robot goals and needs with the desires of the humans that task them with achieving a certain purpose. There are many issues in terms of defining the implicit cultural and ethical context of the purposes that are assigned. Consequently, very close attention will be paid to these problems with the aim of extracting insights that help to advance in the solution of the alignment problem within the field of robotics.
Pillar is working with the IMOL community to establish links and synergies with robotics and AI stakeholders across Europe. What would you say to European industry players and research groups who would like to know more about the project’s goals? Do you see a path to strengthen cooperation and produce meaningful results in the years to come?
PILLAR-Robots has a very strong collaborative character and, even though it’s a very research-oriented project, its aims are to produce results that can be further developed into products and strategies to advance the field of industrial and service robotics. Thus, we have defined a dissemination strategy that segments its scope into three different groups of stakeholders: General Stakeholders (AI, data, robotics ecosystem and society in general), Friendly stakeholders (potential Associated Partners, technology and industrial organizations), and Associated Partners (industrial players and SMEs interested in the project results). The last two categories seek to involve industrial partners, especially SMEs in the development and evaluation of results of the project so that a critical mass of stakeholders can be gathered to guarantee this continuity towards final products and services. In this line, we have organized and are organizing different workshops within the relevant research community, such as the IMOL 2023 workshop in Paris last September, as well as in industrial robotics forums. As an example, we can cite the forthcoming workshop at ERF 2024 on Mobile Manipulation we are organizing, as a chance to strengthen synergies with industry, sister projects, and other like-minded initiatives.
With a commitment to lifelong open-ended learning and a keen focus on aligning robot goals with human desires, Pillar-Robots is poised to revolutionize industries, redefine human-robot relationships, and usher in a new era where robots are not just tools but trusted companions in our journey of progress.