Carnegie Mellon University Robotics Institute Research Guide

Graphics

The current computer graphics group at RI consists of six (6) faculty members, bolstered recently by faculty hires with research thrusts in the areas of computer graphics architectures and high speed rendering, and computer graphics and computer vision. Over the past five years (2005-2010), the Robotics Institute has also broadened the graphics group in the areas of simulation of very large systems (fluids, crowds, molecules) and cross-disciplinary areas in graphics and vision (augmented reality and animation).

The graphics group maintains a focus on data-driven computer graphics which exploits vast amounts of data from the real world, as well as a focus in techniques that allow a user to have intuitive, real-time interaction with characters and scenes. Recent projects include:

• Realistic representation of character skin deformations, dynamic motions, trips, grasps, and clothing, driven from captured human motion data and human subjects studies. Many of these techniques also allow real-time user control and/or real-time editing [Hodgins, Pollard, and Treuille].
• Realistic renderings of fluids and separation of global vs. direct illumination properties of scenes, driven by sampling those scenes under controlled lighting conditions [Narasimhan],
• New interpretation and interface tools for photographs, as well as new views of scenes (e.g., see-through views), driven from very large collections of images and video [Efros, Narasimhan, and Sheikh],
• A new understanding of problem solving strategies for difficult problems such as protein folding, obtained from tremendous numbers of samples of users engaged in game play [Treuille].

We offer a full complement of graphics courses, maintaining the goal of offering at least one and typically two advanced undergraduate courses and at least one graduate course per semester. Advanced undergraduate courses include Computational Photography, Technical Animation, Animation Art and Technology, and Vision Sensors.

Collaborations between computer graphics and other groups within CMU are very strong, as each member of the graphics group spans multiple research areas. Hodgins and Pollard are active in humanoid robotics, Efros, Narasimhan, and Sheikh are active in computer vision, and Treuille is active in computational biology. The connection between groups in graphics and vision has recently been reinforced by the move of many of the graphics and vision faculty and students into a new joint space on the second floor of Smith Hall.

One major addition to our local collaborative environment is the founding of Disney Research Pittsburgh, directed by Jessica Hodgins, which has brought a number of talented graphics researchers right next to campus. Collaborations between graphics faculty and other local institutions are further facilitated through our adjunct faculty Iain Matthews and Katsu Yamane at Disney Research Pittsburgh as well as Liz Marai at the University of Pittsburgh.

Continue Reading: Humanoids