Using collision costs to maintain safe distances for autonomous vehicles
Aspects of the disclosure provide a method of controlling an autonomous vehicle. The method may include receiving sensor data identifying a road user in a lane adjacent to the autonomous vehicle. A minimum magnitude of road user behavior for the road user that is likely to result in a collision with the autonomous vehicle. A collision cost may be determined based on the determined minimum magnitude of road user behavior and a probability density function of previously observed magnitudes of road user behavior. The autonomous vehicle may be controlled based on the collision cost.
1 . A method of controlling an autonomous vehicle, the method comprising:
receiving, by one or more processors, sensor data identifying a road user in a lane adjacent to the autonomous vehicle;
determining, by the one or more processors, a minimum magnitude of road user behavior for the road user that is likely to result in a collision with the autonomous vehicle;
determining, by the one or more processors, a collision cost based on the determined minimum magnitude of road user behavior and a probability density function of previously observed magnitudes of road user behavior; and
controlling, by the one or more processors, the autonomous vehicle based on the collision cost.
2 . The method of claim 1 , wherein the determined minimum magnitude of road user behavior is a minimum amount of deceleration.
3 . The method of claim 2 , wherein the collision cost goes to zero as the minimum amount of deceleration goes towards a maximum possible braking value.
4 . The method of claim 1 , wherein the determined minimum magnitude of road user behavior is a minimum amount of lateral movement.
5 . The method of claim 1 , wherein determining the collision cost includes integrating over deceleration values and time.
6 . The method of claim 1 , further comprising, prior to determining the collision cost:
identifying a characteristic based on the sensor data; and
selecting the probability density function from a plurality of probability density functions based on the characteristic.
7 . The method of claim 6 , wherein the characteristic is a speed of the road user.
8 . The method of claim 6 , wherein the characteristic is traffic density.
9 . The method of claim 6 , wherein the characteristic is an observed behavior of the road user.
10 . The method of claim 6 , wherein the characteristic is an observed behavior of a second road user.
11 . The method of claim 1 , further comprising, prior to determining the collision cost:
identifying a characteristic based on map information; and
selecting the probability density function from a plurality of probability density functions based on the characteristic.
12 . The method of claim 11 , wherein the characteristic is the autonomous vehicle is approaching a curve.
13 . The method of claim 11 , wherein the characteristic is the autonomous vehicle is approaching an intersection.
14 . The method of claim 1 , further comprising, using the collision cost to generate a trajectory, and wherein controlling the autonomous vehicle is based on the generated trajectory.
15 . The method of claim 1 , further comprising, using the collision cost to select one of a plurality of generated trajectories, and wherein controlling the autonomous vehicle is based on the selected trajectory.
16 . The method of claim 1 , wherein the determined minimum magnitude of road user behavior is determined for a situation in which the autonomous vehicle was to move into an adjacent lane in which the road user is located.
17 . The method of claim 1 , wherein the determined minimum magnitude of road user behavior is determined for a situation in which the road user was to move laterally towards the autonomous vehicle.
18 . A system for controlling an autonomous vehicle, the system comprising one or more processors configured to:
receive sensor data identifying a road user in a lane adjacent to the autonomous vehicle;
determine a minimum magnitude of road user behavior for the road user that is likely to result in a collision with the autonomous vehicle if the autonomous vehicle were to move behind the road user;
determine a collision cost based on the determined minimum magnitude of road user behavior and a probability density function of previously observed lead agent deceleration magnitudes; and
control the autonomous vehicle based on the collision cost.
19 . The system of claim 18 , herein the one or more processors are further configured to determine the minimum magnitude of road user behavior that is likely to result in a collision with the autonomous vehicle if the autonomous vehicle were to move into an adjacent lane in which the road user is located.
20 . The system of claim 18 , further comprising the autonomous vehicle.