Overview of the project

Recent years have witnessed a profound transformation in the use of intelligent technology in our daily lives. This technology is now an ubiquitous reality that provides us with personal assistants deployed in several platforms, ranging from small devices, such as smart phones, smart watches and smart speakers, to domestic robots, such as Roomba and Pepper.

As the number of robots continues to increase, we expect to see robots interacting more and more with a variety of other robots and humans. In many of these interactions, the robots may share the same goal and consequently cooperate to achieve that goal. However, many robots may not have coordination and standard communication protocols, because they have been designed by different developers and at different times. Hence, robots have to observe the teammates, reason about their strategy and act in order to achieve the common goal.

This area of research is called ad hoc teamwork and is focused on building learning agents, such as softbots or robots, that engage in cooperative tasks with other unknown agents. The agent must effectively coordinate with the other agents towards completion of the intended task, not relying on any predefined coordination strategy. In this project, we focus on the challenging goal of ad hoc teams of humans and robots. The project aims to go beyond the typical “master-slave” and “one-robot-one-human” type of interaction, pushing step changes in the current state-of-the-art in terms of human-robot interactions.

Goal and Objectives

The two main project goals are to explore the scientific and technological challenges involved in:

• Developing human-robot ad hoc teams. We will test the developed technology in the context of a physical game involving humans and robots in order to investigate how to extend ad hoc teamwork methods and multi-agent planning under uncertainty to cooperative scenarios involving human and robot agents.

• Developing robust forms of natural interaction, including natural language interactions, and how contextual and environment information, collected by a sensor network, can improve the interaction between robot and human agents during joint collaborative tasks involving activities in a physical environment.

---

Team

The project is a collaboration between INESC-ID and PUC-Rio.

INESC-ID

• José Alberto Sardinha
  Principal Investigator
• Francisco S. Melo
  Senior Researcher
• João Ribeiro
  PhD Student

PUC-Rio

• Ruy Luiz Milidiú
  Principal Investigator
• Júlio Duarte
  Senior Reseacher
• Rafael Rocha
  Senior Reseacher

Publications

---