Techniques for navigating an autonomous vehicle based on perceived risk

Techniques for navigating an autonomous vehicle based on perceived risk

A system and method of selecting a trajectory for a vehicle operating in an autonomous mode, comprising the steps of obtaining 1502 a path, a set of lateral constraints, and a set of speed constraints, wherein the lateral constraints define first and second distances between a first and second side of the vehicle and first and second obstacles and the speed constraints define a maximum or minimum speed of the vehicle; selecting 1504 a trajectory by determining 1506 that the trajectory satisfies the lateral and speed constraints and determining 1508 that the trajectory minimises an overall cost, the overall cost including first and second cost components representing a perceived risk to the first and second sides of the vehicle; and navigating 1510 the vehicle according to the trajectory.

Bidirectional path optimization in a grid

Method of navigating an autonomous driverless vehicle, by sensing the environment 800 surrounding the ego vehicle 508 and populating an occupancy map (fig 6c), generating a provisional reference path 804 to a target destination location 808, and determining an actual travel trajectory to intermediate interval points 802, the intermediate points 802 being proximal to points 806 along the reference path 804. The actual travel trajectory is determined based on the reference path and a travel direction of the vehicle (e.g. forward or reversing backward). Multiple potential travel paths, each comprising series of unit-distance steps, may be generated and one selected based on e.g. collision risk, comfort cost, or degree of difference in heading. The position of the vehicle 508 and other obstructing obstacle objects, the reference path 804, and the actual travel trajectory may be converted to the coordinates of the occupancy map (fig 6c) for planning, and returned back to the physical coordinates ...