APE: Active Prototyping Environment

Goal

Geometry

• Evaluation: In such a high speed, real-time, system geometric evaluation can quickly become a bottle neck. Methods are required to evaluate models that will produce the evaluation point, surface tangents, surface normal, and second derivative information.
• Representation: During tracing, the contact point slides along the surface boundary of the model. The model is made up of multiple surfaces connected by trimming data structures. Efficient computation requires special data structures that can store the model compactly, but in a form easily queried.
• Global Computation: Some information required is global in scope. For instance, the future point of contact is always the global closest point. There are currently no real-time methods for finding this point so hybrid real-time approaches are investigated.
• Local Computation: During tracing, local closest points are required that shadow the users movement along the models surface boundary. Further, these local closest points also supplement the global computations.

Graphics

• Speed: The display of the environment needs to be kept at an interactive rate. We use optimized OGL but are looking into other display methods as well. There is a lot of work into point clouds for example. Somewhat faster than showing polygons in some cases.
• Stereo: We also need to have proper display characteristics to produced head tracked stereoscopic displays. This greatly aids the sense of immersion within the environment.

Concurrent Processing

• Multi-machine: The viewer can be run on one machine while each individual device controller can be run on its own machine.
• Multi-process: In the case of the Phantom controller there are two distinct processes: one that talks to the APE viewer, and one that talks to the Phantom. Shared memory is used to allow the two to communicate with each other.
• Multi-thread: Everything is multi-threaded. The viewer has a thread for the graphics and for each of the devices it communicates with. Each device has a thread to do its communication with APE as well as the device. Also, there are various other threads to do assorted global tasks, like tracking a globally closest point.

Communication

• Networking: In order to talk across multiple machines, as well as multiple architectures, a machine independent network protocol is required.
• Shared memory: Fast communication between processes.
• Zero-wait buffering: Data needs to be both written and read as fast as possible. Avoiding any locking is a huge win. We use an approach that allows zero wait on both read and write while also providing the most recent data to readers.

Participants

• Thomas V Thompson II
• David Johnson
• Elaine Cohen

Publications

• Thompson II, T.V., and Cohen, E., "Direct Haptic Rendering of Complex Trimmed NURBS Models," Proc. 8th Annual Symp. on Haptic Interfaces for Virtual Environment and Teleoperator Systems, (Nashville, TN), ASME, November 14-16, 1999.
  [Postscript]
• Johnson, D.E., Thompson II, T.V., Kaplan, M., Nelson, D., and Cohen, E., "Painting Textures with a Haptic Interface," in Proc. Virtual Reality '99, (Houston, TX), pp. 282-285, IEEE, March 13-17, 1999.
  [PDF]
• Johnson, David E., and Cohen, Elaine, "An Improved Method for Haptic Tracing of Sculptured Surfaces," Symp. Haptic Interfaces, Proc. ASME Dynamic Systems and Control Division, DSC-Vol. 64, Anaheim, CA, Nov. 15-20, 1998, pp. 243-248. [pdf]
• Thompson II, T.V., Nelson, D.D., Cohen, E., and Hollerbach, J.M., "Manueverable Models Within A Haptic Virtual Environment," in Proc. 6th Annual Symp. on Haptic Interfaces for Virtual Environment and Teleoperator Systems, (Dallas, TX), pp. 37-44, ASME, Nov. 15-21, 1997.
  [Postscript]
• Thompson II, T.V., Johnson, D.E., and Cohen, E., "Direct Haptic Rendering Of Sculptured Models," in Proc. Symposium on Interactive 3D Graphics, (Providence, RI), pp. 167-176, ACM, April, 1997.
  [Postscript] [PDF]

Support