Figure 1 - uploaded by Pierre E Dupont
Content may be subject to copyright.
Schilling Robotic Systems Titan II manipulator, one of the leading commercial robots for undersea applications. (a) Remote manipulator arm is hydraulically powered. (b) The passive master arm provides no force feedback.
Source publication
This paper presents results on a teleoperator expert assistant - a system that in cooperation with a human operator estimates properties of remote environment objects in order to improve task performance. Specifically, an undersea connector-mating task is investigated in the laboratory using a PHANToM master and WAM remote manipulator. Estimates of...
Context in source publication
Context 1
... the clearance between the connector and receptacle is small, so a few degrees of angular misalignment can cause the connector to become jammed [6]. Second, sensory feedback is limited: because of stringent cost and reliability requirements, the master control device does not provide force feedback (Figure 1). Visual information is also restricted to monocular cameras that may be obstructed by sediment from nearby drilling operations. ...
Citations
... Shared autonomy has appeared in many problem domains, including remote telepresense [11,21,41], assistive robotic manipulation [29,34,44], and assistive navigation [4,20]. In shared autonomy, one of the most persistent challenges has been correctly identifying the pilot's intentions or goals. ...
... Shared autonomy has appeared in many problem domains, including remote telepresense [12,22,41], assistive robotic manipulation [30,35,44], and assistive navigation [5,21]. In shared autonomy, one of the most persistent challenges has been correctly identifying the pilot's intentions or goals. ...
Shared autonomy is an operational concept in which a user and an autonomous agent collaboratively control a robotic system. It provides a number of advantages over the extremes of full-teleoperation and full-autonomy in many settings. Traditional approaches to shared autonomy rely on knowledge of the environment dynamics, a discrete space of user goals that is known a priori, or knowledge of the user's policy -- assumptions that are unrealistic in many domains. Recent works relax some of these assumptions by formulating shared autonomy with model-free deep reinforcement learning (RL). In particular, they no longer need knowledge of the goal space (e.g., that the goals are discrete or constrained) or environment dynamics. However, they need knowledge of a task-specific reward function to train the policy. Unfortunately, such reward specification can be a difficult and brittle process. On top of that, the formulations inherently rely on human-in-the-loop training, and that necessitates them to prepare a policy that mimics users' behavior. In this paper, we present a new approach to shared autonomy that employs a modulation of the forward and reverse diffusion process of diffusion models. Our approach does not assume known environment dynamics or the space of user goals, and in contrast to previous work, it does not require any reward feedback, nor does it require access to the user's policy during training. Instead, our framework learns a distribution over a space of desired behaviors. It then employs a diffusion model to translate the user's actions to a sample from this distribution. Crucially, we show that it is possible to carry out this process in a manner that preserves the user's control authority. We evaluate our framework on a series of challenging continuous control tasks, and analyze its ability to effectively correct user actions while maintaining their autonomy.
... There is a long history of work in shared autonomy across a variety of domains including remote telepresence [14,21,42], assistive robotic manipulation [28,36,43], and assistive navigation [8,20]. Early work assumes that the user's goal is known [13,29,42,51] to the agent. ...
... There is a long history of work in shared autonomy across a variety of domains including remote telepresence [14,21,42], assistive robotic manipulation [28,36,43], and assistive navigation [8,20]. Early work assumes that the user's goal is known [13,29,42,51] to the agent. ...
Shared autonomy provides an effective framework for human-robot collaboration that takes advantage of the complementary strengths of humans and robots to achieve common goals. Many existing approaches to shared autonomy make restrictive assumptions that the goal space, environment dynamics, or human policy are known a priori, or are limited to discrete action spaces, preventing those methods from scaling to complicated real world environments. We propose a model-free, residual policy learning algorithm for shared autonomy that alleviates the need for these assumptions. Our agents are trained to minimally adjust the human's actions such that a set of goal-agnostic constraints are satisfied. We test our method in two continuous control environments: LunarLander, a 2D flight control domain, and a 6-DOF quadrotor reaching task. In experiments with human and surrogate pilots, our method significantly improves task performance even though the agent has no explicit or implicit knowledge of the human's goal. These results highlight the ability of model-free deep reinforcement learning to realize assistive agents suited to complicated continuous control settings with minimal knowledge of user intent.
... Prior research has investigated shared autonomy and assistive teleoperation in an effort to make robot teleoperation easier, safer, and more efficient. Early shared autonomy systems assumed that the robot would be given information about the current user goal or desired behavior, which was then used to determine assistance strategies that helped users accomplish their predefined goal [1,5,11]. While systems relying on such knowledge can be efficient, we are interested in creating more fluid interactions that do not require the overhead of explicitly defining user goals. ...
Robotic teleoperation can be a complex task due to factors such as high degree-of-freedom manipulators, operator inexperience, and limited operator situational awareness. To reduce teleoperation complexity, researchers have developed the shared autonomy control paradigm that involves joint control of a robot by a human user and an autonomous control system. We introduce the concept of active learning into shared autonomy by developing a method for systems to leverage information gathering: minimizing the system's uncertainty about user goals by moving to information-rich states to observe user input. We create a framework for balancing information gathering actions, which help the system gain information about user goals, with goal-oriented actions, which move the robot towards the goal the system has inferred from the user. We conduct an evaluation within the context of users who are multitasking that compares pure teleoperation with two forms of shared autonomy: our balanced system and a traditional goal-oriented system. Our results show significant improvements for both shared autonomy systems over pure teleoperation in terms of belief convergence about the user's goal and task completion speed and reveal trade-offs across shared autonomy strategies that may inform future investigations in this space.
... Hence, ensuring the ability to anticipate the needs and goals of each other from behavior during collaborative work is critical to achieve good team performance [5,23]. It is not uncommon in human-robot interaction and assistive teleoperation studies that the robot is assumed to know the human intention [24][25][26][27][28][29][30][31][32]. In some other studies, it is assumed that the human is following one of a predefined goals or paths, and then a classifier is used to decide the human goal [33][34][35][36][37][38][39]. ...
Human-in-the-loop robot control systems naturally provide the means for synergistic human–robot collaboration through control sharing. The expectation in such a system is that the strengths of each partner are combined to achieve a task performance higher than that can be achieved by the individual partners alone. However, there is no general established rule to ensure a synergistic partnership. In particular, it is not well studied how humans adapt to a nonstationary robot partner whose behavior may change in response to human actions. If the human is not given the choice to turn on or off the control sharing, the robot–human system can even be unstable depending on how the shared control is implemented. In this paper, we instantiate a human–robot shared control system with the “ball balancing task,” where a ball must be brought to a desired position on a tray held by the robot partner. The experimental setup is used to assess the effectiveness of the system and to find out the differences in human sensorimotor learning when the robot is a control sharing partner, as opposed to being a passive teleoperated robot. The results of the four-day 20-subject experiments conducted show that 1) after a short human learning phase, task execution performance is significantly improved when both human and robot are in charge. Moreover, 2) even though the subjects are not instructed about the role of the robot, they do learn faster despite the nonstationary behavior of the robot caused by the goal estimation mechanism built in.
... The level of autonomy of the assistive system can be defined as a fixed or a variable entity. In fixed assistance systems, the tasks are distributed between the operator and the autonomous controller and cannot be re-assigned [16] whereas in variable assistance systems the task assignment is changing throughout the task execution [17], [18]. This variation is mostly pre-determined and does not take into account the 'confidence' of the autonomous component in achieving the task at hand. ...
... In addition, the information gain from each executed trajectory (defined in (16)) was analyzed in this experiment as well to check its effectiveness as a token in deciding whether to aggregate the newly executed trajectory to the learning dataset or not. Fig. 6 recounts the behavior of the indicator over time showing a decrease in the information gain with every iteration. ...
Shared control is a key technology for various robotic applications in which a robotic system and a human
operator are meant to collaborate efficiently. In order to achieve efficient task execution in shared control, it is essential to predict the desired behavior for a given situation or context to simplify the control task for the human operator. To do this prediction, we use Learning from Demonstration (LfD), which is a popular approach for transferring human skills to robots. We encode the demonstrated behavior as trajectory distributions
and generalize the learned distributions to new situations. The goal of this paper is to present a shared control framework that uses learned expert distributions to gain more autonomy. Our approach controls the balance between the controller’s autonomy and the human preference based on the distributions of the demonstrated trajectories. Moreover, the learned distributions are autonomously refined from collaborative task executions, resulting in a master-slave system with increasing autonomy that requires less user input with an increasing
number of task executions. We experimentally validated that our shared control approach enables efficient task executions. Moreover, the conducted experiments demonstrated that the developed system improves its performances through interactive task executions with our shared control.
... Mechatronic systems are, consequently, only means for optimizing particular features, such as reducing tremors or improving precision. [8][9][10][11][12][13][14][15][16] This, for example, is the case in innovative devices used for minimally invasive surgery or rehabilitation 17 where surgeons or users replicate their movements in the working area. These tools can be, for example, haptic interfaces with force feedback control [18][19][20][21][22][23][24] or simply joysticks and/or space balls. ...
Human management of robots in many specific industrial activities has long been imperative, due to the elevated levels of complexity involved, which can only be overcome through long and wasteful preprogrammed activities. The shared control approach is one of the most emergent procedures that can compensate and optimally couple human smartness with the high precision and productivity characteristic to mechatronic systems. To explore and to exploit this approach in the industrial field, an innovative shared control algorithm was elaborated, designed and validated in a specific case study. © SAGE Publications Ltd, unless otherwise noted. Manuscript content on this site is licensed under Creative Commons Licenses.
... The history of assistive intelligence for teleoperation begins in 1963, with the work of Goertz (Goertz, 1963) on using teleoperated manipulators for handling radioactive material. Since then, research on this topic has proposed a great variety of methods for assistance, ranging from the robot having full control over all or some aspect of the motion (Rosenberg, 1993; Marayong, Li, Okamura, & Hager, 2003; Debus, Stoll, Howe, & Dupont, 2000; You & Hauser, 2011; Hauser, 2012; D.-J. Kim et al., 2012; Marayong, Okamura, & Bettini, 2002; Demiris & Hayes, 2002; Fagg, Rosenstein, Platt, & Grupen, 2004), to taking control (or releasing it) at some trigger (Kofman, W., Luu, & Verma, 2005; Shen, Ibanez-Guzman, Ng, & Chew, 2004; Anderson, Peters, Iagnemma, & Overholt, 2010), to never fully taking control (Crandall & Goodrich, 2002; Aigner & McCarragher, 1997; You & Hauser, 2011; Marayong et al., 2002; Aarno, Ekvall, & Kragic, 2005). ...
... Kim et al., 2012; Marayong, Okamura, & Bettini, 2002; Demiris & Hayes, 2002; Fagg, Rosenstein, Platt, & Grupen, 2004), to taking control (or releasing it) at some trigger (Kofman, W., Luu, & Verma, 2005; Shen, Ibanez-Guzman, Ng, & Chew, 2004; Anderson, Peters, Iagnemma, & Overholt, 2010), to never fully taking control (Crandall & Goodrich, 2002; Aigner & McCarragher, 1997; You & Hauser, 2011; Marayong et al., 2002; Aarno, Ekvall, & Kragic, 2005). For example, (Debus et al., 2000) propose that the robot should be in full control of the orientation of a cylinder while the user is inserting it into a socket. In (Kofman et al., 2005), the robot takes over to complete the grasp when close enough to the target. ...
n this paper, we explore a class of teleoperation problems where a user controls a sophisticated device (e.g. a robot) via an interface to perform a complex task. Teleoperation interfaces are funda- mentally limited by the indirectness of the process, by the fact that the user is not physically execut- ing the task. In this work, we study intelligent and customizable interfaces: these are interfaces that mediate the consequences of indirectness and make teleoperation more seamless. They are intelli- gent in that they take advantage of the robot’s autonomous capabilities and assist in accomplishing the task. They are customizable in that they enable the users to adapt the retargetting function which maps their input onto the robot. Our studies support the advantages of such interfaces, but also point out the challenges they bring. We make three key observations. First, although assistance can greatly improve teleperation, the decision on how to provide assistance must be contextual. It must depend, for example, on the robot’s confidence in its prediction of the user’s intent. Second, although users do have the ability to provide intent-expressive input that simplifies the robot’s pre- diction task, this ability can be hindered by kinematic differences between themselves and the robot. And third, although interface customization is important, it must be robust to poor examples from the user.
... Since then, research on this topic has proposed a great variety of methods for assistance, ranging from the robot having full control over all or some aspect of the motion [1]- [6], [10], [11], to taking control (or releasing it) at some trigger [9], [13], [20], to never fully taking control [4], [6]- [8], [14]. For example, Debus et al. [3] propose that the robot should be in full control of the orientation of a cylinder while the user is inserting it into a socket. In [9], the robot takes over to complete the grasp when close enough to the target. ...
In assistive teleoperation, the robot helps the user accomplish the desired task, making teleoperation easier and more seamless. Rather than simply executing the user's input, which is hindered by the inadequacies of the interface, the robot attempts to predict the user's intent, and assists in ac-complishing it. In this work, we are interested in the scientific underpinnings of assistance: we formalize assistance under the general framework of policy blending, show how previous work methods instantiate this formalism, and provide a principled analysis of its main components: prediction of user intent and its arbitration with the user input. We define the prediction problem, with foundations in Inverse Reinforcement Learning, discuss simplifying assumptions that make it tractable, and test these on data from users teleoperating a robotic manipulator under various circumstances. We propose that arbitration should be moderated by the confidence in the prediction. Our user study analyzes the effect of the arbitration type, together with the prediction correctness and the task difficulty, on the performance of assistance and the preferences of users.
