Modeling for Semi-Autonomous Control of Cooperative Vehicles
Abstract
This report details work completed in an AFOSR program focused on semi-autonomous control of cooperative vehicles. Coupled operator-multiple vehicle systems are modeled in a unified framework using probabilistic graphs to yield a methodology for analyzing semi-autonomous systems. The framework uses conditional probabilistic dependencies between elements, leading to a Bayesian network with probabilistic evaluation capability. Both discrete and continuous human decisions are modeled using statistical tools such as softmax and discrete, and Parzen and Gaussian sum distributions. Statistical formalism is maintained in order to enable probabilistic analysis and prediction. The theory has been applied to human decision data using RoboFlag, a multiple robot simulator of capture the flag; data was collected in a series of three tests, jointly developed and implemented with AFRL/HECP. Program contributions summarized here include: (1) probabilistic modeling of both discrete and continuous human decisions using probabilistic graphs; (2) Coupled human-vehicle probabilistic models; (3) empirical studies of human-vehicle interaction with a focus on adaptive tasking; (4) theory and experiments with formal human sensor networks; (5) cooperative geolocation using uninhabited aerial vehicles.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 15, 2009
- Accession Number
- ADA505176
Entities
People
- Mark E. Campbell
Organizations
- Cornell University