Plan-Based Control of Joint Human-Robot Activities

Abstract

Cognition in technical systems is especially relevant for the interaction with humans. We present a newly emerging application for autonomous robots: companion robots that are not merely machines performing tasks for humans, but assistants that achieve joint goals with humans. This collaborative aspect entails specific challenges for AI and robotics. In this article, we describe several planning and action-related problems for human-robot collaboration and point out the challenges to implement cognitive robot assistants. MotivationResearch on intelligent autonomous robots has mainly focused on machines that perform certain services for humans, such as robot butlers or autonomous cars [7,16,24,26]. The interaction in such systems is restricted to a user interface to give a task to the robot and when acting, such a robot has to ensure the safety of humans in its vicinity.A new class of robot applications takes a different view: a robot assistant performs tasks together with a human in order to achieve a joint goal. This approach combines the cognitive capabilities and manual dexterity of humans with the strength, endurance and accuracy of robots. Rosenthal et al. [20] use the term "symbiotic relationship" to describe the reciprocal complementation of skills.On the one hand, the combination of human and robot skills moves useful robot applications closer to the state of marketable products. Whereas a robot butler would be expected to prepare a cup of tea all by itself, a robot assistant might only be expected to bring the kettle and the tea for a person to prepare the tea herself. On the other hand, the close interaction and collaboration with humans requires a range of new cognitive capabilities for robots in order to coordinate the human's and robot's actions in a way that feels natural to people. This paper is concerned with the action side of humanrobot collaboration. This includes AI techniques for planning and plan execution as well as specific action primitives for navigation and joint manipulation tasks. Application ScenariosTo illustrate the idea of assistant robots, we present two application scenarios where a collaboration of a human and a robot can be of substantial help, even if the robot has restricted capabilities.

The role of roles: Physical cooperation between humans and robots

“…In this article, we assume shared goals in terms of known intermediate configurations of the manipulated object. A path for the cooperating dyad can be precomputed by the robot from planning as proposed by Kirsch et al (2010) or from human demonstration (Medina et al, 2011).…”

Rapid Prototyping of Planning, Learning and Control in Physical Human-Robot Interaction

“…a more pro-active robot reduces the required human effort to solve the task. Our proof-of-concept studies show that planbased [14] as well as motion reproduction strategies [20,21] can successfully generate goal-oriented behaviors in full-scale cooperative transport tasks and improve assistance. These results expose complementary properties such as adaptation to human preferences in case of learning approaches and explicit consideration of environment knowledge in planning-based approaches indicating the potential for synergies.…”

“…With our rapid prototyping experimental system, different alternative implementations of each layer are experimentally evaluated: Joint planning through automatic segmentation and verbal communication in combination with intention recognition implements perception and modeling of the task and the partner behavior [20]. Motion generated from the learned human motion model and the estimated human intention or plan-based alternatives [14] implement the generative layer. Two alternatives are further investigated in this work.…”

Predictability in human-agent cooperation

“…[17]). In particular, the use of human behavioural models providing insights into the human nature from the psychological point of view is desired [3,12]. These models form the theoretical basis for predicting human behaviour and can provide information about the personality, habits and capabilities of humans, e.g.…”