Exploring Control Authority Preferences in Robotic Arm Assistance for Power Wheelchair Users.

This paper uses mixed methods to explore the preliminary design of control authority preferences for an Assistive Robotic Manipulator (ARM). To familiarize users with an intelligent robotic arm, we perform two kitchen task iterations: one with user-initiated software autonomy (predefined autonomous actions) and one with manual control. Then, we introduce a third scenario, enabling users to choose between manual control and system delegation throughout the task. Results showed that, while manually switching modes and controlling the arm via joystick had a higher mental workload, participants still preferred full joystick control. Thematic analysis indicates manual control offered greater freedom and sense of accomplishment. Participants reacted positively to the idea of an interactive assistive system. Users did not want to ask the system to only assist, by taking over for certain actions, but also asked for situational feedback (e.g., 'How close am I (the gripper)?', 'Is the lid centered over the jug?'). This speaks to a future assistive system that ensures the user feels like they drive the system for the entirety of the task and provides action collaboration in addition to more granular situational awareness feedback.

The Role of Adaptation in Collective Human-AI Teaming.

This paper explores a framework for defining artificial intelligence (AI) that adapts to individuals within a group, and discusses the technical challenges for collaborative AI systems that must work with different human partners. Collaborative AI is not one-size-fits-all, and thus AI systems must tune their output based on each human partner's needs and abilities. For example, when communicating with a partner, an AI should consider how prepared their partner is to receive and correctly interpret the information they are receiving. Forgoing such individual considerations may adversely impact the partner's mental state and proficiency. On the other hand, successfully adapting to each person's (or team member's) behavior and abilities can yield performance benefits for the human-AI team. Under this framework, an AI teammate adapts to human partners by first learning components of the human's decision-making process and then updating its own behaviors to positively influence the ongoing collaboration. This paper explains the role of this AI adaptation formalism in dyadic human-AI interactions and examines its application through a case study in a simulated navigation domain.

Machine Teaching for Human Inverse Reinforcement Learning.

As robots continue to acquire useful skills, their ability to teach their expertise will provide humans the two-fold benefit of learning from robots and collaborating fluently with them. For example, robot tutors could teach handwriting to individual students and delivery robots could convey their navigation conventions to better coordinate with nearby human workers. Because humans naturally communicate their behaviors through selective demonstrations, and comprehend others' through reasoning that resembles inverse reinforcement learning (IRL), we propose a method of teaching humans based on demonstrations that are informative for IRL. But unlike prior work that optimizes solely for IRL, this paper incorporates various human teaching strategies (e.g. scaffolding, simplicity, pattern discovery, and testing) to better accommodate human learners. We assess our method with user studies and find that our measure of test difficulty corresponds well with human performance and confidence, and also find that favoring simplicity and pattern discovery increases human performance on difficult tests. However, we did not find a strong effect for our method of scaffolding, revealing shortcomings that indicate clear directions for future work.

Detection of Subtle Context-Dependent Model Inaccuracies in High-Dimensional Robot Domains.

Autonomous robots often rely on models of their sensing and actions for intelligent decision making. However, when operating in unconstrained environments, the complexity of the world makes it infeasible to create models that are accurate in every situation. This article addresses the problem of using potentially large and high-dimensional sets of robot execution data to detect situations in which a robot model is inaccurate-that is, detecting context-dependent model inaccuracies in a high-dimensional context space. To find inaccuracies tractably, the robot conducts an informed search through low-dimensional projections of execution data to find parametric Regions of Inaccurate Modeling (RIMs). Empirical evidence from two robot domains shows that this approach significantly enhances the detection power of existing RIM-detection algorithms in high-dimensional spaces.