DTA

Digital Theses Archive

 

Tesi etd-08312018-152559

Type of thesis
Perfezionamento
Author
SINGH, HARMEET
URN
etd-08312018-152559
Title
Human-Robot Physical Interaction Sensing based on Forced Vibration
Scientific disciplinary sector
ING-IND/34
Course
INGEGNERIA - Biorobotics
Committee
relatore Prof. CIPRIANI, CHRISTIAN
Keywords
  • Human Robot Interaction
  • Machine learning
  • Neural networks
  • Rehabilitation
  • Sensor
  • Vibration
Exam session start date
;
Availability
parziale
Abstract
The field Human-Robot Interaction (HRI) includes the study dedicated to understanding, designing, and evaluating robotic systems for use by or with humans. The interaction can be in an industrial or domestic environment. Within this framework object handover between robots and humans is among the fundamental collaborative tasks that should occur in a fluent yet safe manner. The handover between humans are smooth and effortlessly, but human-robot handovers are unintuitive and often unsatisfactory. In this thesis, I propose a completely novel method that helps the robot to detect different events during robot-to-human handover by means of forced vibrations. In the chapter 1, I presented an overview of the state of the art of the methods used to detect the grasp in a robot-to-human handover. In particular, I have shown the sensors used in the detection. I also discussed how robots are expected to assist human in various daily tasks and helps in rehabilitation after some disorder occur in the human body. In chapter 2, I have studied the handover between humans to understand the role of tactile feedback and visual feedback in modulation of the release force implemented by the passer during the exchange of objects. Here, we show that passers use visual-feedback based anticipatory control to trigger the beginning of the release, to launch the appropriate motor program, and adapt such predictions to different speeds of the receiver's reaching out movements. In particular, this study suggest that the passer starts releasing the object in synchrony with the collision with the receiver, regardless of the receiver's speed, but the peak rate of the passer grip force release is correlated with receiver speed. When visual feedback is removed, the beginning of the passer's release is delayed proportionally with the receiver's reaching out speed; however, the correlation between the passer's peak rates of change of grip force is maintained. In chapter 3, I presented a new method for a robotic passer, that is potentially able to estimate the interaction forces by the receiver on the object, thus to achieve fluent and safe handovers. The system proposed uses a vibrator that energies the object and an accelerometer that monitors the vibration propagated through the object, which changes with handover events/states. By using this approach, conventional force/torque sensor could be replaced. To proof this concept, we designed an experimental setup and assessed the viability of the approach offline using machine-learning techniques as the decision methods. Finally, the potential of the system was also assessed online using the same setup interacting with humans. In chapter 4, reports another real-life application of the proposed approach in the field of rehabilitation robotics. For this, I designed the thimble, which is the one of the part of DESC (Discrete Event-Driven Sensory Feedback Control) - glove. DESC- glove is a device that send the feedback to the user through the arm cuff attached to the user's arm by elastic band. Arm cuff consists of a Printed circuit board, a battery and vibrator that sense the signal from the thimble. My thimble consisted of an accelerometer and combined with a vibrator and used the same approach of forced vibration, in this case on the thimble and compared with the thimble instrumented with a piezo film. The main idea was to compare our approach with an existing sensor. We presented the sensitivity in terms of the force applied on the thimbles at different positions with different force rate.
Files