Embedded Humans Project
Unmanned vehicle platforms are becoming increasingly ubiquitous, providing persistence, precision, survivability, flexibility, and safety for a number of diverse and demanding operations. Further, we are called upon to guarantee that human decision makers embedded in the midst of either fully or partially automated systems can execute missions with variability in the environment and their cognitive abilities. Thus, we need to design systems that blend human cognitive understanding and control capability with autonomy in vehicle ensembles.
We have gathered specialists in control and decision theory, cognitive science, AI, neuroscience, robotics, and human-machine interfaces from University of California, Berkeley, UCLA and Stanford University to develop a fundamentally new science of “human centered automation” or “embedded humans”.
We have organized this MURI research around the following four technical thrust areas.
Thrust 1 - Modeling of mixed initiative systems: in this thrust of the project, we study how to incorporate the human’s cognitive resources and limits into the design of the control architecture in the very early stages of design specification.
Thrust 2 - Content-based data acquisition and representation: to ensure accurate modeling and representation of human knowledge and to understand/execute human requests and responses in an autonomous way, we address efficient extraction of information in the signals or data gathered from the users and their surrounding cyber-physical environments or finding relevant information and knowledge in a massive database.
Thrust 3 - Dynamic assignment of commands to distributed resources: we develop computationally efficient and robust methods for automatic allocation and re-allocation of tasks in order to decompose team objectives to individual primitives and to define team reconfigurations in dynamic environments.
Thrust 4 - Provably correct mixed initiative systems: in this thrust, we address one of the most challenging problems in mixed initiative systems, which is deriving proofs of correctness and tools for the synthesis of such systems. These can be probabilistic statements of either safety or liveness properties, but can include clauses about the limitations of the algorithms, so that they know when to ask the human for direction.
Our team consists of experts in control and decision theory, cognitive science, artificial intelligence, neuroscience, robotics, and human-machine interfaces from the three institutions of University of California at Berkeley, University of California at Los Angeles and Stanford University.
Dr. Shankar Sastry (Electrical Engineering and Computer Sciences Department, UC Berkeley)
Dr. Ruzena Bajcsy (Electrical Engineering and Computer Sciences Department)
Dr. Francesco Borrelli (Mechanical Engineering Department)
Dr. Tom Griffiths (Psychology and Cognitive Science Department)
Dr. Karl Hedrick (Mechanical Engineering Department)
Dr. Stuart Russell (Electrical Engineering and Computer Sciences Department)
Dr. Claire Tomlin (Electrical Engineering and Computer Sciences Department)
Dr. Leo Guibas (Computer Science Department)
Dr. Surya Ganguli (Applied Physics Department)
Dr. Noah Goodman (Psychology Department)
Dr. Stefano Soatto (Computer Science Department)
Dr. Ehsan Elhamifar (EECS Department, UC Berkeley)
Dr. Siddharth Srivastava (EECS Department, UC Berkeley)
Dr. Lei Li (EECS Department, UC Berkeley)
Dr. William Cushing (EECS Department, UC Berkeley)
Nathan Lam, Dorsa Sadigh, Elena L. Carano, Theresa Lin, Chang Liu, Robert Matthew, Jingming Dong, Jared Garvey, Shih-Yuan Liu, Yusuf Erol, Theresa Lin, Joshua Abbott, Jessica Hamrick
MURI annual meeting: UC Berkeley, September 17, 2014.
MURI summer meeting: UC Berkeley, June 20, 2014. Meeting Schedule
Research meeting: Theresa Lin: Modeling Human Behavior in a Dynamic Task, March 2014.
Research meeting: Dr. Siddharth Srivastava: Open-Universe POMDPs, February 2014.
Research meeting: Dr. Ehsan Elhamifar: Information Extraction via Convex Optimization, November 2013.
Research meeting: Dr. Allen Yang: Sparse and Low-Rank Representation for Biometrics, October 2013.
MURI annual meeting: UC Berkeley, May 22, 2013.
To modify this page, use Subversion.