PREdicting MObility using STATistics (PreMoStat)

The proposed work addresses the problem of determining whether a given small unmanned ground vehicle (SUGV) can traverse a given terrain, when the state of both the SUGV and the terrain are not known exactly. In our previous work, we developed computational models of an Army-relevant tracked SUGV-the PackBot-traversing terrain. Using SolidWorks, we built three-dimensional solid models of the vehicle and terrain features. Using ADAMS, we simulated the PackBot negotiating the terrain. The results of the simulations agreed well with real-world experiments. Phase I of PreMoStat builds directly on the previous and will develop a Monte Carlo method for predicting off-road mobility, and compare the predicted and actual mobility on step, ditch, and slope obstacles. Task 1, Mobility Prediction, structures our existing software for mobility prediction, and adds tools for visualization and measuring computational resource usage. Task 2, Statistical Mobility Prediction, applies Monte Carlo analysis, and explores of higher-risk statistical approaches. Task 3, Comparison of Predicted and Actual Mobility, runs a real robot on real terrain. Task 4, Project Management, includes all of the non-technical efforts needed to complete Phase I and prepare for Phase II.