Intelligent Path Prediction for Vehicular Travel

The goal location g is assumed to be from the set {g1,g2,...,gm}. The cost criterion c is assumed to be from the set {c1,c2,...,cn}. Research in the field of intelligent planning assists in the synthesis of candidate cost criteria; for example, the methods of Donald and Xavier (1988), Kanayama and DeHaan (1988), Lee and Preparata (1984), Mitchell (1986, 1988), Sharir and Schorr (1986), and Suh and Shin (1988) investigate path planning with the cost criteria of distance, safety, visibility, and time. A candidate mission objective (g,c) describes a transit mission decision-making strategy. In intelligent path prediction, one seeks the particular candidate mission objective that best explains the observed motion. After establishing the predicted mission objective, the future path leading to the goal location is predicted. An optimal path from any point in the environment to a particular goal location g can be obtained by propagating a reverse search based on a cost criterion c from the goal location g; Dijkstra’s algorithm is applicable (Dijkstra 1959; Hart, Nilsson, and Raphael 1968; Mitchell 1986, 1988; Nilsson 1980). The search pointers indicate the optimal solution paths from any location within the search frontier to the goal location g. Reverse-search results maintained in a gradient field representation have utility in path-planning applications for mobile robots (Payton 1990); in this dissertation, reverse-search results provide useful information for the purpose of path prediction. A gradient field representation of a route plan provides, for each location in the map, the best direction to travThe problem of predicting the motion of a vehicle has been investigated by several researchers. Many have used Kalman filter techniques based on the equations of vehicle motion; these techniques most accurately predict shortterm motion. In contrast, my dissertation (Krozel 1992)1 presents a methodology for intelligent path prediction, where predicting the motion of an observed vehicle is performed by reasoning about the decision-making strategy of the vehicle’s operator. With intelligent path prediction, the long-term mission objective of the vehicle is being predicted in addition to the short-term motion. Thus, when applied to predicting the motion of a car, an intelligent predictor will attempt to predict the final destination—say, for example, the vehicle appears to be going to the post office or the art museum—in addition to predicting which streets will be used. The theory is also applicable to predicting air vehicle travel, so that for a military application, the target (from a set of plausible targets) and the threat-avoidance policy (from a set of plausible policies), in addition to the route, can be predicted.