101 of the Greatest Hits
The following is a representative list of research achievements compiled by
Patrick H. Winston. Some of the listed items date back to the founding of
the laboratory during the 1960s, but the list is skewed toward work done
during the past five years.
- ACTOR model of parallel computation
- Albedo maps and their role in satellite image understanding
- Algorithms for massively parallel image processing
- Analogy- and precedent-based learning
- Bit-map displays
- Calculation of shape from stereo information
- Canny edge detector
- Change detection and navigation using optical flow
- Computation of shape from shading
- Computational complexity theory of symbolic planning
- Configuration space and its role in path planning and grasping
- Connection machine---first computer with thousands of processors
- Conspiracy-numbers approach to game playing
- Constraint propagation and its role in vision and in general reasoning
- Constraint suspension and its role in debugging digital hardware
- Dependency-directed backtracking and truth maintenance
- Designed behavior heuristic and its role in testing digital hardware
- Diagram understanding in systems for automatic analysis of dynamical systems
- Diagram understanding systems for reasoning about mechanical systems
- Diaphanography, or the use of scattered light for medical imaging
- Digital orrery and determination of instability of Pluto
- EMACS, one of the world's most influential text editors
- Envisioning approach to robot planning
- Exaggeration as a means integrating qualitative and quantitative reasoning
- Explanation based learning of form-function relationship, as in cup example
- Extremely low voltage interconnection circuitry
- FLAVORS system, a predecessor to the Common Lisp Object System
- Frames and inheritance ideas
- Friction viewed as an aid to manipulation
- FRL, a precursor to the frames component of many expert system shells
- Function sharing heuristic and its role in mechanical design
- Ghenghis-to-Cog line of experimental robots
- Guaranteed reliable assembly in uncertain environments
- Highly articulated, multifingered hands for use in haptics research
- Intelligent room as a substitute for display-keyboard-mouse interaction
- Interdigitated syntactic/semantic activity in natural language processing
- Jelly Bean machine, a high-performance massively parallel machine
- Knowledge acquisition and learning through family resemblance
- Learning of English phonological pluralization rules
- Linear combinations of two-dimensional images in animation
- Linear combinations of two-dimensional images in visual recognition
- Liquid crystal-on-silicon technology for eyeglass mounted displays
- LISP language for symbolic manipulation
- LISP Machine
- LOGO language for teaching principles of computing to children
- Machines that walk, hop, run, turn somersaults, and jump through hoops
- Magnebots, or herds of microrobots driven by surrounding magnetic fields
- Markov random fields for multimodal visual fusion
- Marr's principles of research methodology
- Memory intensive arm control algorithms that learn to perform graceful motions
- Minsky's formulation of steps toward Artificial Intelligence
- Mobile telecomputing architecture for computing infrastructure
- Models for work in organizations
- Multiple-arm robot planning
- Natural language interfaces to databases and command activation
- Near-miss, single-shot learning, as in arch example
- NETL, a connectionist memory model
- Perceptron convergence theorem and other proofs of capabilities and limits
- PLANNER language and independent invention of ideas in PROLOG
- Polymer muscle-like actuators powered by acid-base solutions
- Principle-based parsing
- Qualitative reasoning about physical events
- Radial-basis-function approach to neural nets
- RAF approach to object recognition using model-based constrained search
- Recursive Lagrangian dynamics algorithm for arm control
- Registration of images to models by maximizing reflectance consistency
- Regularity discovery in protein databases
- Regularization theory applied to vision problems
- Repetition as a means for learning visual routines
- SCHEME, a language for teaching programming principles
- Scientific and engineering problem solving systems
- Serial-elastic robotic actuators
- Shape determination from motion information
- Signal description experiments contributing to the development of scale space
- Smart gunk concept for embedding microprocessors in paint and materials
- Smart structures strengthened via deflection sensing and corrective actuation
- Smoothed local symmetries for object description and recognition
- Society of Mind as a theory of intelligence
- START natural language system and annotation-based information access
- Streams and counterstreams model of visual recognition
- Subsumption architecture as a theory of intelligence
- Surface interpolation from sparse visual data
- Symbolic integration and groundwork for MACSYMA
- Tactile sensors for robot manipulation
- Telepresence devices that enable applications such as surgical rehearsal
- The primal sketch and the 2 1/2 D sketch as objectives for
visual information processing
- Theorem proving by selective attention to relevant properties
- Transfer-frame approach to metaphor
- TRANSIT architecture for parallelizing almost-sequential programs
- Transition space, a change-centered language for reasoning about
events and causality
- Variable-precision logic and reasoning
- Vibration control through staged activation
- Visual inspection of VLSI chips
- Visual routines as a mechanism for coupling high-level goals to
low-level image processing
- VLSI chips for difference of Gaussian and other special purpose vision computations
- VLSI design using rule based systems and other AI inspired ideas
- Wait-and-see model for language acquisition
- Wait-and-see model for language parsing
- Wall-sized displays using overlapped, limited-quality projectors
- X-ray vision for surgeons via multimodal registration algorithms
- Zero-crossing technique for edge detection
