Final Report Summary - TRAVERSE (Towards Very Large Scale Human-Robot Synergy)
Project Website: http://users.isr.ist.utl.pt/~aahmad/traverse/doku.php(se abrirá en una nueva ventana)
* A summary description of the project objectives:
Distributed multi-robot teams consisting of a large number of robots operating in close cooperation with humans is fundamental for a variety of scenarios where robotics and automation technologies are being employed increasingly. The aim of this project was to study how to achieve maximum synergy between teams of robots and humans, without jeopardizing human safety and comfort, within the constraints of resources, e.g. computational capacity of the robots, sensor and actuator costs.
The first core objective of this project was to develop cooperative methods for robot team perception functionalities that are scalable to a very large number of robots.
The second core objective was to model and study human and system factors that affect joint task efficiency when human operators interact with large scale robot teams in order to perform a collaborative task. These factors could include human abilities to reason about space, their memory, ease of interaction with the multi-robot system, etc.
* A description of work performed since the beginning of the project.
Concerning the first objective
-- A highly scalable method for multirobot cooperative perception (CP) was developed. CP in this context included robot-self localization, teammate localization and target tracking.
-- A model-based approach for person-detection using an RGBD sensor was developed and implemented in a real robot scenario.
-- A novel method for cooperative stereo vision-based target tracker was developed and implemented on a team of two aerial robots.
Concerning the second objective
-- An experimental framework was conceptualized and developed to study human-multirobot interaction (H-MRI). The joint task in this study was survivor search mission in a disaster-hit scenario. Using this novel experimental framework, an elaborate psychophysical study was performed on 35 human subjects. Various human and system factors were studied within the context of scalable CP.
* A description of the main results achieved:
Concerning the first objective
-- The scalable multirobot cooperative perception (CP) method was implemented and verified on two separate datasets. While the first dataset included a challenging scenario of multiple ground-based robots tracking a soccer ball in a highly dynamic environment, the second dataset included multiple ground-based robots tracking each other at a much lower frequency (than the previous dataset) making the dataset even more challenging. The CP method included an optimization-based estimator that not only runs in real-time and is scalable to a large number of robots but is also asymptotically stable and convergent.
Concerning the second objective
-- Through extensive experiments on a large number of human subjects (35) we found the role of system and human factors on the collaborative task efficiency in an H-MRI search mission scenario within the context of cooperative perception. The first main finding is that the collaborative task is performed much better in a scenario where the robots are fully autonomous in performing exploration of the world while the human operators are tasked only with searching for survivors in the explored regions of the world. When human operators are additionally tasked with controlling the robots, they perform significantly poorer unless they possess good video gaming experience or training.
-- Our second main finding was that there is only a slight benefit in increasing the number of robots in the team performing the collaborative task under the assumption that a single human operator is responsible for the actual survivor search and classification task.
-- We also found that for a small number of robots there is no significant difference in the collaborative task efficiency between a fully controlled multirobot team (where each robot is individually commanded) or a formation controlled team (where the group is commanded as a whole).
* Potential impact and use:
-- Primarily, our scalable CP method is a completely novel approach of estimation where we have explored combining the filter and optimization-based estimation techniques to exploit the benefits of both while overcoming each of their shortcomings. This should open the door for further research in such techniques, specially for obtaining scalability in estimation techniques.
-- Secondly, our findings related to human and system factors in the joint task efficiency of a collaborative human-multirobot team will directly impact human-robot interaction guidelines for developing an effective architecture for human-multirobot search missions. To the best of our understanding, this is a first such project that has studied these factors under the scenario of scalable cooperative perception.
* A summary description of the project objectives:
Distributed multi-robot teams consisting of a large number of robots operating in close cooperation with humans is fundamental for a variety of scenarios where robotics and automation technologies are being employed increasingly. The aim of this project was to study how to achieve maximum synergy between teams of robots and humans, without jeopardizing human safety and comfort, within the constraints of resources, e.g. computational capacity of the robots, sensor and actuator costs.
The first core objective of this project was to develop cooperative methods for robot team perception functionalities that are scalable to a very large number of robots.
The second core objective was to model and study human and system factors that affect joint task efficiency when human operators interact with large scale robot teams in order to perform a collaborative task. These factors could include human abilities to reason about space, their memory, ease of interaction with the multi-robot system, etc.
* A description of work performed since the beginning of the project.
Concerning the first objective
-- A highly scalable method for multirobot cooperative perception (CP) was developed. CP in this context included robot-self localization, teammate localization and target tracking.
-- A model-based approach for person-detection using an RGBD sensor was developed and implemented in a real robot scenario.
-- A novel method for cooperative stereo vision-based target tracker was developed and implemented on a team of two aerial robots.
Concerning the second objective
-- An experimental framework was conceptualized and developed to study human-multirobot interaction (H-MRI). The joint task in this study was survivor search mission in a disaster-hit scenario. Using this novel experimental framework, an elaborate psychophysical study was performed on 35 human subjects. Various human and system factors were studied within the context of scalable CP.
* A description of the main results achieved:
Concerning the first objective
-- The scalable multirobot cooperative perception (CP) method was implemented and verified on two separate datasets. While the first dataset included a challenging scenario of multiple ground-based robots tracking a soccer ball in a highly dynamic environment, the second dataset included multiple ground-based robots tracking each other at a much lower frequency (than the previous dataset) making the dataset even more challenging. The CP method included an optimization-based estimator that not only runs in real-time and is scalable to a large number of robots but is also asymptotically stable and convergent.
Concerning the second objective
-- Through extensive experiments on a large number of human subjects (35) we found the role of system and human factors on the collaborative task efficiency in an H-MRI search mission scenario within the context of cooperative perception. The first main finding is that the collaborative task is performed much better in a scenario where the robots are fully autonomous in performing exploration of the world while the human operators are tasked only with searching for survivors in the explored regions of the world. When human operators are additionally tasked with controlling the robots, they perform significantly poorer unless they possess good video gaming experience or training.
-- Our second main finding was that there is only a slight benefit in increasing the number of robots in the team performing the collaborative task under the assumption that a single human operator is responsible for the actual survivor search and classification task.
-- We also found that for a small number of robots there is no significant difference in the collaborative task efficiency between a fully controlled multirobot team (where each robot is individually commanded) or a formation controlled team (where the group is commanded as a whole).
* Potential impact and use:
-- Primarily, our scalable CP method is a completely novel approach of estimation where we have explored combining the filter and optimization-based estimation techniques to exploit the benefits of both while overcoming each of their shortcomings. This should open the door for further research in such techniques, specially for obtaining scalability in estimation techniques.
-- Secondly, our findings related to human and system factors in the joint task efficiency of a collaborative human-multirobot team will directly impact human-robot interaction guidelines for developing an effective architecture for human-multirobot search missions. To the best of our understanding, this is a first such project that has studied these factors under the scenario of scalable cooperative perception.