A Review of Communicating Robot Learning during Human-Robot Interaction (2312.00948v1)

Abstract: For robots to seamlessly interact with humans, we first need to make sure that humans and robots understand one another. Diverse algorithms have been developed to enable robots to learn from humans (i.e., transferring information from humans to robots). In parallel, visual, haptic, and auditory communication interfaces have been designed to convey the robot's internal state to the human (i.e., transferring information from robots to humans). Prior research often separates these two directions of information transfer, and focuses primarily on either learning algorithms or communication interfaces. By contrast, in this review we take an interdisciplinary approach to identify common themes and emerging trends that close the loop between learning and communication. Specifically, we survey state-of-the-art methods and outcomes for communicating a robot's learning back to the human teacher during human-robot interaction. This discussion connects human-in-the-loop learning methods and explainable robot learning with multi-modal feedback systems and measures of human-robot interaction. We find that -- when learning and communication are developed together -- the resulting closed-loop system can lead to improved human teaching, increased human trust, and human-robot co-adaptation. The paper includes a perspective on several of the interdisciplinary research themes and open questions that could advance how future robots communicate their learning to everyday operators. Finally, we implement a selection of the reviewed methods in a case study where participants kinesthetically teach a robot arm. This case study documents and tests an integrated approach for learning in ways that can be communicated, conveying this learning across multi-modal interfaces, and measuring the resulting changes in human and robot behavior. See videos of our case study here: https://youtu.be/EXfQctqFzWs

Understanding the Communication Loop in Human-Robot Interaction

The Importance of Communication

To realize effective human-robot collaboration, there must be a transparent flow of information between humans and robots. Researchers have developed numerous algorithms for robots to learn from human input. In parallel, communication interfaces have been designed to transmit the internal state of the robot back to the human facilitator. Prior studies often treat robot learning and communication independently, focusing either on the algorithms or the interfaces, rather than the integration of both. In recent times, the trend is shifting towards an interdisciplinary framework where both learning and communication are seen as a continuous loop.

Learning Algorithms and Communication

Robots acquire knowledge from human guidance and feedback, and this process can now be communicated back to human operators using two mechanisms: implicit and explicit communication. Implicit communication happens naturally during human-in-the-loop learning processes, where the robot reflects its learning progress through its actions. Explicit communication goes further by having the robot convey preprocessed “signals” to provide a more direct explanation of what it has learned. The explicit approach includes rendering waypoints, highlighting critical task states, or even summarizing decision-making trees.

Advancing Communication Interfaces

The rise of immersive technologies like augmented reality (AR) and haptic feedback is revolutionizing how robots communicate their internal states. AR empowers the human user to see virtual overlays in the physical task environment, offering spatial and directional context, while haptic feedback can provide tactile cues on force requirements or alert on uncertainties. Moreover, multisensory interfaces that combine visual, auditory, and haptic feedback are emerging to provide a richer and more intuitive understanding of the robot's learning process.

Measuring Human-Robot Dynamics

To gauge the impact of these communicative methods, researchers employ a mixture of objective performance metrics and subjective assessments of the human operator’s perspective. Objective measurements include the accuracy of the robot's task execution and human teaching quality. Subjectively, surveys assess human factors like trust in the robot, the intuitiveness of interacting with the robot, and the perceived ease of interpreting the robot's feedback.

Outcomes of Closing the Loop

By closing the loop - communicating the robot's internal learning state back to the human - studies have identified significant benefits. Improved teaching methods emerge as the human instructor can target specific areas where the robot displays uncertainty. Trust is fortified when people can predict how the robot will behave. Additionally, humans and robots tend to co-adapt, optimizing collaboration and sometimes leading to role reversals or new behaviors in joint tasks.

Future Research Directions

Despite advances, questions remain about optimizing the translation of complex robot learning into understandable feedback and about the effectiveness of interfaces in conveying robot learning. Moreover, real-time measurement tools are needed to better capture the human’s interpretation of the robot’s state - a step toward achieving mutual understanding between human teachers and robot learners.

Case Study: Interactive Teaching with a Robot Arm

A case paper involving participants kinesthetically teaching a robot arm reinforces these findings. When explicit feedback was given through GUI or AR and haptic interfaces, not only did robot learning improve, but participants also expressed a preference for these methods over implicit communication. This aligns with the notion that explicitly conveying robot learning enhances human teaching and robot performance while fostering a favorable human-robot interaction experience.

In summary, bridging the gap between robot learning and human understanding is pivotal for seamless collaboration. As this field moves forward, the interplay between learning, communication, and measurement is set to define the future of human-robot teamwork.