Michael Patrick Johnson

Integrated learning for interactive synthetic characters

The ability to learn is a potentially compelling and important quality for interactive synthetic characters. To that end, we describe a practical approach to real-time learning for synthetic characters. Our implementation is grounded in the techniques of reinforcement learning and informed by insights from animal training. It simplifies the learning task for characters by (a) enabling them to take advantage of predictable regularities in their world, (b) allowing them to make maximal use of any supervisory signals, and (c) making them easy to train by humans.We built an autonomous animated dog that can be trained with a technique used to train real dogs called clicker training. Capabilities demonstrated include being trained to recognize and use acoustic patterns as cues for actions, as well as to synthesize new actions from novel paths through its motion space.A key contribution of this paper is to demonstrate that by addressing the three problems of state, action, and state-action space discovery at the same time, the solution for each becomes easier. Finally, we articulate heuristics and design principles that make learning practical for synthetic characters.

Sympathetic interfaces: using a plush toy to direct synthetic characters

We introduce the concept of a sympathetic inter

Physically interactive story environments

ace forcontrolling an animated synthetic character in a 3D virtualenvironment. A plush doll embedded with wireless sensors is used tomanipulate the virtual character in an iconic and intentionalmanner. The interface extends from the novel physical input devicethrough interpretation of sensor data to the behavioral brain ofthe virtual character. We discuss the design of the interface andfocus on its latest instantiation in the Swamped! exhibit atSIGGRAPH 98. We also present what we learned from hundreds ofcasual users, who ranged from young children to adults.

Automated creation of visual routines using genetic programming

Most interactive stories, such as hypertext narratives and interactive movies achieve an interactive feel by allowing the user to choose among multiple story paths. In this paper we discuss physically interactive environments with narrative structure in which the ability to choose among multiple story lines is replaced with having users, first, interact with the story characters in small, local windows of the narrative and, second, actively engage their bodies in movement, In particular, we found that compelling interactive narrative story systems can be perceived as highly responsive, engaging, and interactive even when the overall story has a single-path structure, in what we call a less-choice, more-responsiveness approach to the design of story-based interactive environments. We have also observed that unencumbering, rich sensor technology can facilitate user immersion in the experience as the story progresses--users can act as they typically would without worrying about manipulating a computer interface. To support these arguments, the paper describes the physical setup, the interactive story, the technology, and the user experience of four projects developed at the MIT Media Laboratory: KidsRoom, It/I, Personal Aerobics Trainer, and Swamped!

Out of the box (panel): toys break the screen barrier

Traditional machine vision assumes that the vision system continually recovers a complete, labeled description of the world from the visual field. Many researchers have criticized this model and proposed an alternative model which considers visual perception as a distributed collection of task-specific, context-driven visual routines. Ullmans visual routines model (1984) of intermediate vision describes one way this might be accomplished. To date, most researchers have hand-coded task-specific visual routines for actual implementations of systems requiring simple vision. We propose an alternative approach in which visual routines are created using artificial evolution, a supervised learning approach. We present results from a series of runs on a simple vision problem using real camera data, in which simple Ullman-like visual routines were evolved using genetic programming. Results were accurate and able to generalize.

Method and system for facilitating wireless, full-body, real-time user interaction with a digitally represented visual environment

Transmedia is a new genre of software-based product that only recently has emerged into the marketplace. These applications blur the line between physical play in the real world and virtual play in the digital world. Are these applications precursors to the eventual integration of the computer into daily life? Just as ATM’s broke through a major sociological barrier to widespread acceptance of the computer as an appliance, will these products become the next vanguard of consumer acceptance?