Controlling an autonomous vehicle using cost maps

A method of controlling an autonomous vehicle, the method comprising:

• receiving sensor data;
• generating, based on the received sensor data, and for each area in a first set of areas in an environment through which the autonomous vehicle is moving, (i) first data indicating whether the area is currently occupied, and (ii) second data indicating whether the area is currently expected to be occupied at one or more future instances in time;
• receiving navigation data configured to guide the autonomous vehicle through the environment toward a destination;
• generating a plurality of cost maps based on (i) the first data, (ii) the second data, and (iii) the navigation data, each cost map specifying numerical values representing costs, at a respective instance of time, of the autonomous vehicle occupying each of the first set of areas, wherein
• generating the plurality of cost maps includes generating a first cost map corresponding to a first instance of time,
• generating the first cost map includes generating a first set of costs, of a second set of one or more areas around a detected object, that decrease with distance from the detected object to form a halo around the detected object,
• generating the first set of costs includes generating the first set of costs based in part on a speed, at the first instance of time, of the detected object, such that a size and/or an intensity of the halo depends upon the speed of the detected object, and
• the second set of areas is a subset of the first set of areas;
• generating, based on the plurality of cost maps, a path through the environment;
• and
• generating decisions for maneuvering the autonomous vehicle toward the destination based on the path.

An autonomous vehicle comprising:

• one or more operational subsystems configured to maneuver the autonomous vehicle, wherein the operational subsystems collectively control at least speed and direction of the autonomous vehicle;
• a sensor system including one or more sensor types and configured to generate sensor data;
• and
• a computing system configured to
• receive the sensor data from the sensor system,
• generate, based on the received sensor data, and for each area in a first set of areas in an environment through which the autonomous vehicle is moving, (i) first data indicating whether the area is currently occupied, and (ii) second data indicating whether the area is currently expected to be occupied at one or more future instances in time,
• receive navigation data configured to guide the autonomous vehicle through the environment toward a destination,
• generate a plurality of cost maps based on (i) the first data, (ii) the second data, and (iii) the navigation data, each cost map specifying numerical values representing costs, at a respective instance of time, of the autonomous vehicle occupying each of the first set of areas, wherein
• generating the plurality of cost maps includes generating a first cost map corresponding to a first instance of time,
• generating the first cost map includes generating a first set of costs, of a second set of one or more areas around a detected object, that decrease with distance from the detected object to form a halo around the detected object,
• generating the first set of costs includes generating the first set of costs based in part on a speed, at the first instance of time, of the detected object, such that a size and/or an intensity of the halo depends upon the speed of the detected object, and
• the second set of areas is a subset of the first set of areas,
• generate, based on the plurality of cost maps, a path through the environment,
• generate decisions for maneuvering the autonomous vehicle toward the destination based on the path, and
• provide the generated decisions to the one or more operational subsystems to effectuate maneuvering of the autonomous vehicle in accordance with the generated decisions.