Zoltan Derzsi^1.2 (), Francesco Ceccarini^1,2, Robert Volcic^1,2,3; ¹New York University Abu Dhabi, ²NYUAD Center for Artificial Intelligence and Robotics, ³NYUAD Center for Brain and Health

When humans grasp an object, there are several factors such as the natural grasp axis, force closure and minimum torque that impact the selection of the most appropriate contact points on the surface of the object and lead to natural and effortless grasping actions. For this process to be successful, constraints relative to 3D shape perception need also to be considered. It is indeed known that interactions between shape, material properties and lighting conditions can adversely affect 3D shape constancy. However, studies and models on grasp contact point selection have so far rarely paid attention to the effects of these factors. Here, we addressed this gap by focusing on how contact point selection in grasping might be affected by different lighting conditions. Participants were asked to grasp and lift a real 3D globally convex smoothly curved object created by distorting a sphere with a fractal Brownian motion displacement algorithm and presented at different orientations. Instead of presenting the object under diffuse ambient illumination, the object was illuminated with one directional light (cold-white pin-spotlight, 5000 lux) from one side at a time (top-left vs. top-right). This illumination altered the object's appearance such that the illuminated areas were very well visible, but the non-illuminated areas appeared as a very dark attached shadow in our otherwise dark lab environment, similar in appearance to photographs of asteroids. By considering the contact positions of the thumb and index finger just before lifting the object, we observed that the selection of these contact points changed systematically under variations of illuminant direction. Our results imply that the factors affecting the selection of contact points should necessarily include effects of lighting conditions and possibly material properties as well, as they can often influence the visual appearance of 3D objects.

Acknowledgements: We acknowledge the support of the NYUAD Center for Artificial Intelligence and Robotics and the NYUAD Center for Brain and Health, funded by Tamkeen under the NYUAD Research Institute Awards CG010 and CG012.