ABSTRACT
Driver San Francisco was an open-world driving game set in a fictionalized version of the city of San Francisco. This chapter presents the path planning solution Driver San Francisco used: a three-tier path optimization approach that provided locally optimal paths. It describes the three stages of the process—route finding, mid-level path planning, and low-level path optimization. To control the vehicle in the simple physics simulation, a simple AI path-following module was used. This module was exactly the same as the AI that was used to follow paths in the real game. The AI was fed with information about the current dynamic state of the vehicle and details of the path it was expected to follow. Path optimization allowed the system to deal efficiently with a world composed of highly dynamic obstacles, a world which would have been too costly to analyse using more traditional A*-based searching.
