In a haptic shared control situation, human and automation collaborate in a control task by both exerting forces on a control device. A primary issue in haptic shared control is the tuning of the strength of the haptic feedback. In most cases, this tuning represents a compromise between performance, safety, and operator acceptance. Recently a new design philosophy for a haptic shared control system was presented in [1]. This new design philosophy decouples (human compatible) reference trajectory from a shared controller, thereby allowing for easy adaptation to different references (trajectories). The shared controller has a feedforward and feedback loop, producing two distinctly different torques, which are also perceived (felt) differently by the human, analogous to a teacher who corrects each small mistake you make (feedback) and one who ‘shows you the way’ (feedforward). Recent experiments [2] with the new architecture shown in the diagram above found that the new feedforward (LoHS) component was a dominant factor in reducing conflicts between haptic support and operator, resulting in a different ‘feel’ to the haptic shared controller, and increased acceptance.
By removing the feedforward components, the architecture can be made to mimic conventional single-path haptic support systems. In this project, we seek to further explore the role of the feedforward component, by systematically varying the LoHS in an experiment.

Your Task:

• For each chosen setting to be tested in the experiment, the other parts of the architecture are tuned for optimal performance, so that the tested setting can effectively drive a reference trajectory without human intervention. This needs offline Matlab simulation and programming in DUECA.
• You should find that automatically driving the reference trajectory can be done through multiple different haptic feedback and feedforward settings.
• For a chosen set of two or three LoHS/SoHF settings, choose different TDCA torque fractions (at 100% the TDCA torque follows the reference exactly, at a less percentage the human has to contribute as well), for an empirical experiment to test how humans perceive this guidance. Essentially investigating how different shared control settings affect human-automation conflicts and acceptance

Workgroup	Human-Machine Interaction/Cognitive Robotics 3me
Mentors	ir. S. Barendswaard, dr. ir. M.M. van Paassen, dr.ir.Daan Pool, Prof David Abbink, Prof Max Mulder
Begin date	Mar-Jun 2018
Contact	s.barendswaard@tudelft.nl

Keywords: Haptic Shared Control design, acceptance, conflicts.

References

[1] M. M. Van Paassen, R. Boink, D. A. Abbink, M. Mulder, and M. Mulder, “Four design choices for haptic shared control,” Adv. Aviat.
Psychol., pp. 237 – 254, 2017.
[2] W. Scholtens, S. Barendswaard, D.A. Abbink, D.M.Pool, M.M. Van Paassen, ‘A New Haptic Shared Controller Reducing Steering
Conflicts’ IEEE SMC, 2018, to be published.