2017-2020 PhD: Collaborative Hands-on Training on haptic simulators

Student: Angel LICONA

Keywords:: Haptics, Simulation, Hands-on Training, Dual-User

Supervised with Minh Tu PHAM (director)

Period: started in January 2017, to be defended in 2020

Financed by CONACYT Mexico

Summary :
Medical staffs require continuing hands-on training on ever evolving medical methods. For instance, Minimally Invasive Surgery (MIS) procedures have brought much comfort to the patient but has complicated the task of surgeons as they now manipulate their tools through trocars by way of a 2D camera visualization.
During their education, they usually train on black boxes, cadavers or animals (when available), and more recently passive and active simulators, before training on real patients. It has been proven that computer based haptic training simulators lead to an efficient training for advanced tasks (see [Panait09]).
However, in general, the trainee is alone in front of the simulator and cannot take benefit of a supervised training. Hence, in supervised hands-on training, the trainer takes the hands of the trainee in his own hands in order to guide him and to perform difficult gestures. While the hands' motion is driven by the trainer, they both share the haptic feedback derived from the manipulated tools.
Therefore it is difficult for the trainer to dose his forces and for the trainee to feel the right level of forces to apply.
Dual-user systems permit this : each one directly manipulates a different haptic interface acting as a common fake tool while the real tool is actuated by the slave part of the system. This slave part can also be a virtual tool in a virtual environment. They have been introduced by Nudehi et al. in [Nudehi05] and some variations have been studied in [Ghorbanian13], [Khademian11] and
[Razi14]. The common concept is that the interfaces provide force feedback to both master users (trainer and trainee) according to a dominance factor (alpha in [0,1]). When alpha=1 (resp. 0), the trainer (resp. trainee) has full authority on the trainee's (resp. trainer's) device and on the slave. When 0 < alpha < 1, both users share the slave control with a dominance (over the other user) which is function of alpha. This control authority, shared between both users, is chosen according to their relative level of skills and experience. It determines the extent to which the motion of the slave tool depends on their individual commands.
For four years, the Medical Robotics team of Ampere research laboratory has focused on such dual-user simulators (see [Liu15a] and [Liu15b]). An energetic modeling approach has been successfully used to control this architecture in presence of small delays between the three devices for one degree-of-freedom.
However, to be completely useful, the manipulation must be performed with multiple degrees of freedom. Moreover, one has to take into account that communication may have limitations: low bandwidth, packet drops, disconnections, ... as the trainee and the trainer will likely be located in different locations (resp. university and hospital). At last, medical trainers would appreciate a system which enables an automatic evaluation of the quality of the trainee gestures and their evolution.

The outcomes of this PHD project will consist in developing strategies to overcome the network connection defects and provide by the way a robust architecture enabling an effective hands-on training with its automatic evaluation.
To do so, the applicant will have to enhance and extend the model of the current dual-user training system. The current passivity controller will have to be enhanced to take into account multiple degrees of freedom and some defects such as varying delays and packets drops. He/She will also apply and enhance the gesture analysis methods developed by the team [Cifuentes14] but not yet applied on such a training system. Simulations and then experimentations will have to be conducted in order to validate the stability, the transparency, the robustness and the usability by medical students.