Search Results

A common scenario in engineering design is the availability of several black-box functions that describe an event with different levels of accuracy and evaluation cost. Solely employing the highest fidelity, often the most expensive, black-box function leads to lengthy and costly design cycles. Multi-fidelity modeling improves the efficiency of the design cycle by combining information from a small set of observations of the high-fidelity function and large sets of observations of the low-fidelity, fast-to-evaluate functions. In the context of Bayesian optimization, the most popular multi-fidelity model is the auto-regressive (AR) model, also known as the co-kriging surrogate. The main building block of the AR model is a weighted sum of two Gaussian processes (GPs). Therefore, the AR model is well suited to exploit information generated by sources that present strong linear correlations.

Vehicle navigation in off-road environments is challenging due to terrain uncertainty. Various approaches that account for factors such as terrain trafficability, vehicle dynamics, and energy utilization have been investigated. However, these are not sufficient to ensure safe navigation of optionally manned ground vehicles that are prone to detection using thermal infrared (IR) seekers in combat missions. This work is directed towards the development of a vehicle IR signature aware navigation stack comprised of global and local planner modules to realize safe navigation for optionally manned ground vehicles. The global planner used A* search heuristics designed to find the optimal path that minimizes the vehicle thermal signature metric on the map of terrain’s apparent temperature. The local planner used a model-predictive control (MPC) algorithm to achieve integrated motion planning and control of the vehicle to follow the path waypoints provided by the global planner.