Warning:
This page is
provided for historical and archival purposes
only. While the seminar dates are correct, we offer no
guarantee of informational accuracy or link
validity. Contact information for the speakers, hosts and
seminar committee are certainly out of date.
Member of Research Staff Hughes Research Laboratories, Malibu, CA
With modern intelligence gathering methods, such as with the AWACS or J-STARS aircraft, it is now possible to remotely monitor, in real time, the motion of one or more ground or air vehicles. This data gives past and present locations of vehicles but does not in itself indicate their future path or goal. This talk deals with the prediction of future paths and goals of vehicles observed to be moving over a known environment.
Intelligent path prediction addresses the problem of predicting the path of a vehicle performing a transit mission. Such a mission proceeds from a start location to a goal location guided by an intelligent planning strategy (e.g., minimize distance, minimize visibility, maximize safety, etc.). Given the history of a path from a start location to a current location, the objectives are: 1) to estimate the cost criterion guiding the travel, 2) to predict the goal location (or select a goal location from a set of candidate goal locations), and 3) to predict the future path leading to the predicted goal location. First, a cost criterion explaining the decision-making strategy of the observed vehicle is estimated using a correlation measure comparing the observed path data to optimal path search information. This correlation is expressed in terms of the tolerance epsilon of an epsilon-optimal path. Next, a region of plausible goal locations is predicted assuming that the vehicle will proceed using either optimal decisions or epsilon-optimal decisions in the future. The predicted goal location of the vehicle is determined by selecting the point in the region of plausible goal locations that has the highest heuristic merit, as determined by a proposed ranking system. Finally, from auxiliary search information, the future path is predicted. This problem is generalized to predicting the future path of a point vehicle traveling in an arbitrary dimensional space.
Host: Yangsheng Xu (xu@cs.cmu.edu) Appointment: Lalit Katragadda (lalit@cs.cmu.edu)