When: Monday, May 12, 10:30 a.m.

Where: 3305Newell-Simon Hall

Dean R. Anderson

Thesis Proposal

Abstract:
Autonomous navigation by Unmanned Ground Vehicles (UGVs) in rough terrain is currently a problem of much interest and with many applications. Currently, models of vehicle mobility are being used in a feed-forward fashion to improve performance within planning and control. Such models allow autonomy systems to reason about the consequences of their actions.

However, the mobility models typically deployed are relatively simple and lack robustness to changes in the terrain, environment, or vehicle configuration. This may cause less than optimal behavior by a planning system evaluating predicted paths for cost, such as when a path is predicted to be free of obstacles, but when executed results in a collision.

In this thesis, we propose a system for automatically identifying a vehicle model using conventional, on board state-estimation and terrain perception sensors. Such a system should be able automatically adapt to new or changing environments, vehicle damage or wear and tear. Research areas to be addressed include model structure, convergence and observability.

Furthermore, a natural extension of such a learned model is a principled formulation for propagation of model uncertainties. Confidences for both terrain and vehicle parameters can be directly simulated to provide a probabilistic region of travel along the predicted trajectory. Such information could be leveraged by a planning system to reduce risk and increase performance.

Thesis Committee Members:
Alonzo Kelly, Chair
George Kantor
Carl Wellington
Karl Iagnemma, Massachusetts Institute of Technology

<< Back