A new methodology to evaluate the performance of physics simulation engines in haptic virtual assembly

In this study a new methodology to evaluate the performance of physics simulation engines when used in haptic virtual assembly applications is proposed. This methodology can be used to assess the performance of any physics engine. To prove the feasibility of the proposed methodology, two third party physics simulation engines - Bullet and PhysXtm - were evaluated. Eight assembly tests comprising variable geometric and dynamic complexity were conducted. The strengths and weaknesses of each simulation engine for haptic virtual assembly were identified by measuring different parameters such as task completion time, influence of weight perception and force feedback. The proposed tests have led to the development of a standard methodology by which physics engines can be compared and evaluated. The results have shown that when the assembly comprises complex shapes, Bullet has better performance than PhysX. It was also observed that the assembly time is directly affected by the weight of virtual objects. A more comprehensive study must be carried out in order to evaluate and compare the performance of more physics simulation engines. The influence of collision shape representation algorithms on the performance of haptic assembly must be considered in future analysis. The performance of physics simulation engines in haptic-enabled VR applications had been remained as an unknown issue. The main parameters of physics engines that affect the haptic virtual assembly process have been identified. All the tests performed in this study were carried out with the haptic rendering loop active and the objects manipulated through the haptic device.