Perceived realism of haptic rendering methods for bimanual high force tasks: original and replication study

Realistic haptic feedback is a key for virtual reality applications in order to transition from solely procedural training to motor-skill training. Currently, haptic feedback is mostly used in low-force medical procedures in dentistry, laparoscopy, arthroscopy and alike. However, joint replacement procedures at hip, knee or shoulder, require the simulation of high-forces in order to enable motor-skill training. In this work a prototype of a haptic device capable of delivering double the force (35 N to 70 N) of state-of-the-art devices is used to examine the four most common haptic rendering methods (penalty-, impulse-, constraint-, rigid body-based haptic rendering) in three bimanual tasks (contact, rotation, uniaxial transition with increasing forces from 30 to 60 N) regarding their capabilities to provide a realistic haptic feedback. In order to provide baseline data, a worst-case scenario of a steel/steel interaction was chosen. The participants needed to compare a real steel/steel interaction with a simulated one. In order to substantiate our results, we replicated the study using the same study protocol and experimental setup at another laboratory. The results of the original study and the replication study deliver almost identical results. We found that certain investigated haptic rendering method are likely able to deliver a realistic sensation for bone-cartilage/steel contact but not for steel/steel contact. Whilst no clear best haptic rendering method emerged, penalty-based haptic rendering performed worst. For simulating high force bimanual tasks, we recommend a mixed implementation approach of using impulse-based haptic rendering for simulating contacts and combine it with constraint or rigid body-based haptic rendering for rotational and translational movements.