SCS Undergraduate Thesis Topics
|Akkarit Sangpetch||Dave Touretzky||Visualizing Robot Behavior with Self-Generated Storyboards|
Achieving intelligent behavior on a mobile robot requires a mix of sensory processing, navigation, object manipulation, and human-robot interaction. Behavioral routines are often programmed as a collection of finite state machines, with states corresponding to actions, and state transitions triggered by asynchronous sensory events or timer expirations. Debugging these complex real-time behaviors or making them intelligible to a human observer can be challenging. Development of GUI tools can help programmers simplify this process.
In this project, we develop a graphical "storyboard" representation for visualizing a robot's behavior over time. Such a representation has uses beyond debugging. It can be used to make a robot's behavior comprehensible to its users. And it can provide a visual record of the robot's "performance" on a task, suitable for publication or display. The tool allows user to manipulate state machine diagram and automatically generate a storyboard view, based on layout of the state machine, from robot's execution traces. The tool is implemented in Java on a PC, and communicates with a Sony AIBO robot dog running the Tekkotsu application development framework developed at Carnegie Mellon.