Causal-HRI: Causal Learning for Human-Robot Interaction

Opening and Introduction

Keynote Prof Karinne Ramirez-Amaro

Title: Explainable AI meets Robotics - Robots that Learn Causal-based models to predict and prevent failures.

Abstract: The advances in Collaborative Robots (Cobots) have rapidly increased with the development of novel data- and knowledge-driven methods. These methods allow robots, to some extent, to explain their decisions. This research area is known as Explainable AI and is gaining importance in the robotics community. One advantage of such methods is the increase of human trust towards Cobots since robots can explain their decisions, especially when errors occur or when facing new situations. Explainability is a challenging and important component when deploying Cobots into real and dynamic environments. In this talk, I will introduce our proposed causal-based approach to empower robots with rapid prediction and prevention of both immediate and future failures. I will explain our efforts to automatically learn the cause-effect relationships of tasks, and the generation of explanations for unsuccessful actions. Our goal is to enable robots to understand “why” things happen, which will allow them to learn from their mistakes, thus improving their future performances. Therefore, the new generation of Cobots should reason based on past experiences while providing explanations in case of errors. Thus, improving the autonomy of robots and humans’ trust to work with robots.

Keynote Prof Alison Gopnik

Title: Empowerment as causal learning, causal learning as empowerment.

Abstract: Cognitive scientists have applied advances in our formal understanding of causation in philosophy and computer science, particularly within the Causal Bayes Net formalism, to understand human causal learning. In parallel, in the very different tradition of reinforcement learning, researchers have developed the idea of an intrinsic reward signal called “empowerment”. An agent is rewarded for maximizing the mutual information between its actions and their outcomes, regardless of the external reward value of those outcomes. In other words, the agent is rewarded if variation in an action systematically leads to parallel variation in an outcome so that the value of the action predicts the value of the outcome. The result is an agent that has maximal control over its environment. We argue that “empowerment” may be an important bridge between classical Bayesian causal learning and reinforcement learning and may help to characterize causal learning in humans and enable it in machines. Empowerment may explain distinctive empirical features of children’s causal learning, as well as providing a more tractable computational account of how that learning is possible. More strongly we make the philosophical argument that causal learning and empowerment gain are very closely related. In particular, our hypothesis is that if an agent learns an accurate causal model of the world they will necessarily increase their empowerment, and, vice versa, increasing empowerment will lead to a more accurate (if implicit) causal model of the world. This has implications for both cognitive science accounts of causal learning and for the metaphysics of causation.

Coffee Break

Contributed Talks

• ROS-Causal: A ROS-based Causal Analysis Framework for Human-Robot Interaction Applications [paper]
  Luca Castri, Gloria Beraldo, Sariah Mghames, Marc Hanheide, and Nicola Bellotto
• A Human-Centered Approach for Bootstrapping Causal Graph Creation [paper]
  Minh Q. Tram, Nolan B. Gutierrez, and William J. Beksi
• Human causal reasoning guided autonomous object assembly under uncertainty [paper]
  Semanti Basu, Semir Tatlidil, Moon Hwan Kim, Steven Sloman, and R. Iris Bahar
• Causal Influence Detection for Human Robot Interaction [paper]
  Yi-Shiuan Tung, Himanshu Gupta, Wei Jiang, Bradley Hayes, and Alessandro Roncone

Keynote: Prof Holly Yanco

Title: The Impact of Causal Learning on the Design and Evaluation of Human-Robot Interaction.

Abstract: All robots interact with people in some way. Robots navigating in difficult and dynamic environments often need assistance from human operators or supervisors, either in the form of teleoperation or interventions when the robot's autonomy is not able to handle the current situation. Even in more controlled environments, such as office buildings and manufacturing floors, robots may need help from people. As robots become more autonomous and ubiquitous, people will become bystanders to the systems, with little knowledge of the robot’s intent. As we develop robots that have the ability to explain their actions, how will the design and evaluation of human-robot interaction change? This talk will cover examples from a wide variety of applications, ranging from assistive technology to telepresence to exoskeletons to humanoid robots.

Group Discussion

Closing Remarks