Mulit-scale semi autoomous teleoperation for passion remote handling
Lead Research Organisation:
University College London
Department Name: Computer Science
Abstract
Project Title
Information Representation for Supervision of Semi-Autonomous Robots
Summary
This is a CASE Studentship in collaboration with RACE (Remote Applications in Challenging Environments) within UKAEA (UK Atomic Energy Authority).
Remotely supervising maintenance robots for fusion reactors such as JET (Joint European Torus) currently requires many expert operators to have knowledge of the internal state of the robots in addition to the state of the surrounding environment. The large monetary and temporal expense of training and employing multiple experts for each maintenance robot are not expected to be scalable for proposed fusion reactors such as DEMO (DEMOnstration Power Station), where the demand for maintenance robotics (and therefore supervisors) will be much greater.
During remote maintenance operations with autonomous robots, users must have general situational awareness while also being able to observe low level information to support decision making when autonomy fails. An interface that provides unfettered access to this information is likely to overwhelm a human supervisor, as the amount of information associated with a multi-robot system is large and diverse.
These requirements lead to questions about how the display of information within an interface should change based on the state of the multi robot system, the environment and the supervisor.
Research Questions and Objectives
The primary research question is "How does immersive interface design affect the performance and user experience of a human supervising multiple semi-autonomous robots?".
This research aims to explore novel approaches to immersive interface design which will accommodate greater performance and/or reduced workload for human supervisors of multi-robot systems.
Approach and Novelty
This research aims to answer three main questions through experimentation.
1. How different types of information filtering (displaying or hiding information based on a filter) affect the perceived workload and performance of a teleoperator of multiple semi-autonomous ground robots. The novel contribution of this work is that information filtering of this kind has not previously been evaluated for supervising multiple robots.
2. Investigate how immersive multi-modal interfaces can affect situational awareness and perceived workload when supervising multiple ground-based manipulation robots. This will compare a purely visual immersive interface to an interface with visual, sound and haptic feedback. This is a novel investigation into the effects of a multi-modal immersive interface in the context of multi-robot supervision for robots performing maintenance in a Nuclear environment.
3. Investigate how machine learning can be used to predict which information is most useful to be displayed to a user when supervising multiple ground-based manipulation robots. Investigate how this affects the situational awareness
and perceived workload of the supervisor within an immersive interface. This is a novel investigation into how machine learning can affect the performance of a supervisor in the context of multi-robot supervision.
EPSRC Research Areas that are believed to be most relevant to this project are listed here in order of relevance (1 being most relevant):
1. Human-computer interaction
2. Robotics
3. Graphics and visualisation
4. UK Magnetic Fusion Research Programme
Information Representation for Supervision of Semi-Autonomous Robots
Summary
This is a CASE Studentship in collaboration with RACE (Remote Applications in Challenging Environments) within UKAEA (UK Atomic Energy Authority).
Remotely supervising maintenance robots for fusion reactors such as JET (Joint European Torus) currently requires many expert operators to have knowledge of the internal state of the robots in addition to the state of the surrounding environment. The large monetary and temporal expense of training and employing multiple experts for each maintenance robot are not expected to be scalable for proposed fusion reactors such as DEMO (DEMOnstration Power Station), where the demand for maintenance robotics (and therefore supervisors) will be much greater.
During remote maintenance operations with autonomous robots, users must have general situational awareness while also being able to observe low level information to support decision making when autonomy fails. An interface that provides unfettered access to this information is likely to overwhelm a human supervisor, as the amount of information associated with a multi-robot system is large and diverse.
These requirements lead to questions about how the display of information within an interface should change based on the state of the multi robot system, the environment and the supervisor.
Research Questions and Objectives
The primary research question is "How does immersive interface design affect the performance and user experience of a human supervising multiple semi-autonomous robots?".
This research aims to explore novel approaches to immersive interface design which will accommodate greater performance and/or reduced workload for human supervisors of multi-robot systems.
Approach and Novelty
This research aims to answer three main questions through experimentation.
1. How different types of information filtering (displaying or hiding information based on a filter) affect the perceived workload and performance of a teleoperator of multiple semi-autonomous ground robots. The novel contribution of this work is that information filtering of this kind has not previously been evaluated for supervising multiple robots.
2. Investigate how immersive multi-modal interfaces can affect situational awareness and perceived workload when supervising multiple ground-based manipulation robots. This will compare a purely visual immersive interface to an interface with visual, sound and haptic feedback. This is a novel investigation into the effects of a multi-modal immersive interface in the context of multi-robot supervision for robots performing maintenance in a Nuclear environment.
3. Investigate how machine learning can be used to predict which information is most useful to be displayed to a user when supervising multiple ground-based manipulation robots. Investigate how this affects the situational awareness
and perceived workload of the supervisor within an immersive interface. This is a novel investigation into how machine learning can affect the performance of a supervisor in the context of multi-robot supervision.
EPSRC Research Areas that are believed to be most relevant to this project are listed here in order of relevance (1 being most relevant):
1. Human-computer interaction
2. Robotics
3. Graphics and visualisation
4. UK Magnetic Fusion Research Programme
Publications
| Description | A novel algorithm for autonomous exploration of an unknown volume with a robot arm holding a depth camera was developed and published at the International Conference on Intelligent Robots and Systems 2019 Work is currently being undertaken to evaluate the relationship between interface design for multi-robot supervision and the experience of the supervisor using the interface. |
| Exploitation Route | This novel algorithm could be taken and implimented into future scenarios where autonomous exploration of a volume is required. This would be useful for future fusion reactors, as well as many other sectors where autonomous maintenance and inspection tasks need to be carried out. |
| Sectors | Aerospace, Defence and Marine,Chemicals,Construction,Energy |
| URL | https://ieeexplore.ieee.org/document/8968563 |