CHS: Small: Interactive Haptic Assembly and Docking for 3D Shapes

University Of Connecticut, Storrs CT

Investigators

Abstract

Haptic human-computer interaction mechanisms and systems play a critical role in a variety of engineering and scientific activities that rely on the fundamental task of virtual object assembly, from protein docking, drug design and tele-surgery, to advanced manufacturing, rehabilitation, robotics, teleoperation and consumer applications. One of the key long-standing challenges in developing such practical interactive systems is the lack of a proper formulation of the guidance forces that effectively assist the user in the exploration of the virtual environment, from repulsing collisions to attracting proper contact. A secondary difficulty is that of achieving an efficient implementation that can maintain an acceptable haptic refresh rate. Current state-of-the art solutions to these open problems have been developed for severely restricted classes of shapes and motions, and rely heavily on heuristics that exploit drastic geometric limitations. To address these issues, the PI's goal of this research is to develop a purely geometric model for an artificial energy field that favors spatial relations leading to proper assembly of arbitrarily complex shapes. Project outcomes will lead to effective interaction mechanisms for intelligent human-computer or human-robot systems and will open the doors to the development of generic and fully automated assembly planners while simultaneously unlocking new levels of expression and productivity in activities that rely on interactive assembly tasks in a broad range of industrial, scientific and consumer applications, in domains as diverse as 3D user interfaces, engineering, and medical and assistive technologies. The PI's industrial partnerships will facilitate aggressive and widespread technology transfer. To these ends, the energy function is expressed in terms of a convolution of shape-dependent affinity fields that rely on the novel concept of a space-continuous, well-defined, and robust density function, called the Skeletal Density Functions (SDF), whose sublevel sets in the limit are related to an implicit definition of the medial axis. Importantly, the proposed energy field leads to the first practical and automatic approach to detect key features that contribute to proper alignment or assembly, as well as the geometric constraints required for virtual assembly. Moreover, the proposed approach completely avoids the heuristic recipes and manual intervention that are common to existing methods for haptic assembly. The PI's preliminary results show that this research can unify the two haptic interaction phases of free motion and precision assembly, which are common in current haptic simulations, into a single interaction mode, and suggest a generic and automatic constraint model for the so-called virtual fixtures, with no restrictive assumption on the types of the assembly features and shapes involved.