|Ph.D Student||Luko Liel|
|Subject||Improving Robot Teleoperation through Training of Spatial|
|Department||Department of Industrial Engineering and Management||Supervisor||ASSOCIATE PROF. Avi Parush|
|Full Thesis text|
Robot teleoperation allows the operator to manipulate a robot from a safe distance in hazardous environments such as deep sea and space exploration, bomb disposal, or the very fine robot-assisted surgery. Due to limited visibility and partial spatial cues from the remote environment, teleoperation introduces various challenges. These include depth and visual-spatial perception, orientation and eye-hand coordination and geometric reasoning. Another challenge is mastering the controls to manipulate the remote robot. These challenges entail motoric and cognitive skills, and specifically, spatial cognitive skills such as spatial planning, depth perception, estimation of depth and distance, mental rotation, and perspective taking.
To overcome the challenges in teleoperation and enhance teleoperators’ performance, one must improve the spatial cognitive skills, as they are critical for teleoperation. The main goal of this research was to understand whether spatial cognitive skills could be trained and transferred to performing cognitive tasks and real-world compound tasks within the context of robot teleoperation. To devise a solution for spatial skills acquisition during training, this research offers theoretical foundations for a new approach to training spatial cognitive skills.
The theoretical assumption of this research is that there could be a dissociation between the cognitive and the motoric skills within tasks consisting of those skills. Based on this assumption, it is possible to dissociate the skills in training, train the cognitive skill alone, and then combine them within the overall performance of the task. Implementation is achieved through a cognitive simulator training approach that focuses on spatial cognitive skills and requires cognitive effort alone.
The research program addressed two major questions:
Can spatial cognitive skill be trained and transferred to performing:
1. cognitive tasks within the context of a robot teleoperation task?
2. compound teleoperation task that includes cognitive and motoric components?
Four studies were conducted to address these questions. The first two studies examined targeted cognitive training of perspective taking and mental rotation, and the test of transfer of learning to an analogical robot teleoperation task requiring cognitive effort alone. The third study consisted of a thorough cognitive task analysis of robot-assisted surgery procedure. It delineated the spatial cognitive demands, which were the foundation for the subsequent cognitive training. The final study consisted of targeted cognitive training of several spatial skills, and the test of far transfer of learning to a full robot-assisted surgery procedure with both spatial motoric and cognitive tasks. Results showed that targeted cognitive training of spatial cognitive skills improved teleoperation performance in terms of the surgical performance, in a real-world compound task.
Together, the studies contribute the following theoretical and practical implications: 1. the motoric and cognitive skills can be dissociated in a teleoperation task; 2. training spatial cognitive skills can be done with cognitive effort alone; and 3. Targeted cognitive training can improve operator’s performance in a teleoperation task, which requires compound skills.