Rapid Prototyping of Parallel Robot Vision Systems Using Virtual Reality and Systems Simulation

An NSF Institutional Infrastructure Grant (No. CDA-94-01142)

Principal Investigators

The Idea

Summary

The emerging behavioral and active vision paradigms hold that machines that work in the real world must be built and tested in the real world, and that real-world interaction is necessary for perception. Until recently there has been no practical alternative to building hardware in order to test ideas. Unfortunately, working with sophisticated physical manipulators or navigators is an extremely expensive, time-consuming, and sometimes hazardous process. Recent advances in simulation technology in two specific areas have crossed a threshold, enabling us to solve this problem, and to achieve enormous speedup in the design of complex, real-world robotic systems.

The first technology is the simulation of sensory interaction with physical environments (popularly known as virtual reality), which can replace the real world in testing and debugging a system. The second is execution-driven simulation of complex parallel algorithms at the level of individual messages and memory accesses, which can address the performance and low-level real-time problems of interacting processes.

With NSF Funding, we are equipping a laboratory that breaks logically into two halves. One half of the lab, the robotics laboratory, is for building working systems in the real world. The other half, the virtual reality laboratory, is for prototyping and experimentation in the virtual world. This hardware and software is configured so that the computational engines running the robot control algorithms can be swapped transparently between the real-world devices and the virtual devices. In addition, the virtual-world lab contains hardware and software for simulating and analyzing the control algorithms at a low level, where all the timings and complexities of interaction among the various subsystems can be instrumented and tracked.


For more detailed information, including pointers to publications, see the following.


Chris Brown <brown>

Back to Brown's Home Page

Back to URCS Home Page