Sensor and environmental evaluation for vehicles
Aspects of the disclosure provide for controlling behaviors of autonomous vehicles based on evaluation of sensors of those vehicles. For instance, sensor data including distance and intensity information for a point in an environment of an autonomous vehicle may be received. An expected intensity value from a pre-stored fair weather reference map may be identified based on a location of the point. An effective detection range for the sensor may be dynamically determined based on the expected intensity and the intensity information for the point. A behavior of the autonomous vehicle may be controlled based on the effective detection range.
1 . A method comprising:
receiving, by one or more processors, sensor data generated by a LIDAR sensor of a perception system of an autonomous vehicle, the sensor data including distance and intensity information for a point in an environment of the autonomous vehicle;
identifying, by the one or more processors, an expected intensity at a location of the point from a pre-stored fair weather reference map;
dynamically determining, by the one or more processors, an effective detection range of the LIDAR sensor by comparing the expected intensity at the location of the point to the intensity information for the point, wherein the effective detection range corresponds to a distance at which the LIDAR sensor detects an object and identifies one or more characteristics of the object according to one or more threshold minimums, wherein the one or more threshold minimums include a percentage recall on a Lambertian target; and
controlling, by the one or more processors, a behavior of the autonomous vehicle based on the effective detection range.
2 . The method of claim 1 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces of buildings.
3 . The method of claim 2 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces which are generally vertical.
4 . The method of claim 2 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces which are planar.
5 . The method of claim 2 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces which are generally Lambertian.
6 . The method of claim 1 , further comprising:
determining an aperture transmission value based on the expected intensity and the intensity information for the point;
comparing the aperture transmission value to a threshold value; and
based on the comparison, activating a cleaning process for the LIDAR sensor.
7 . The method of claim 1 , further comprising, determining a meteorological optical range value for the point using the expected intensity and the intensity information for the point, and wherein determining the effective detection range includes determining the effective detection range using the meteorological optical range value.
8 . The method of claim 7 , further comprising generating a meteorological optical range map based on the meteorological optical range value and at least one other meteorological optical range value, and wherein controlling the behavior of the autonomous vehicle is further based on the meteorological optical range map.
9 . The method of claim 1 , further comprising, determining an updated effective detection range based on updated sensor data.
10 . The method of claim 1 , wherein the Lambertian target is a Lambertian black target.
11 . The method of claim 1 , wherein the percentage recall on the Lambertian target is 50% recall on a 5% Lambertian black target.
12 . A system comprising one or more processors configured to:
receive sensor data generated by a LIDAR sensor of a perception system of an autonomous vehicle, the sensor data including a distance for a point in an environment of the autonomous vehicle and intensity information for the point;
identify an expected intensity at a location of the point from a pre-stored fair weather reference map;
determine a meteorological optical range value for the point using the expected intensity at the location of the point and the intensity information for the point, the meteorological optical range value being a distance representative of an effect of one or more atmospheric conditions on the LIDAR sensor;
dynamically determine an effective detection range for the LIDAR sensor based on:
a comparison of the expected intensity to the intensity information for the point, wherein the effective detection range corresponds to a distance at which the LIDAR sensor detects an object and identifies one or more characteristics of the object according to one or more threshold minimums; and
the meteorological optical range value; and
control a behavior of the autonomous vehicle based on the effective detection range.
13 . The system of claim 12 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces of buildings.
14 . The system of claim 13 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces which are generally vertical.
15 . The system of claim 13 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces which are planar.
16 . The system of claim 13 , wherein the pre-stored fair weather reference map includes surfels corresponding to surfaces which are generally Lambertian.
17 . The system of claim 12 , wherein the one or more processors are further configured to
determine an aperture transmission value based on the expected intensity and the intensity information for the point;
compare the aperture transmission value to a threshold value; and
control the behavior by activating a cleaning process for the LIDAR sensor based on the comparison.
18 . The system of claim 12 , wherein the one or more processors are further configured to generate a meteorological optical range map based on the meteorological optical range value and at least one other meteorological optical range value, and wherein controlling the behavior of the autonomous vehicle is further based on the meteorological optical range map.
19 . A non-transitory, tangible, computer-readable medium on which instructions are stored, the instructions, when executed by one or more processors, cause the one or more processors to implement a method, the method comprising:
receiving sensor data generated by a LIDAR sensor of a perception system of an autonomous vehicle, the sensor data including distance and intensity information for a point in an environment of the autonomous vehicle;
identifying an expected intensity at a location of the point from a pre-stored fair weather reference map;
dynamically determining an effective detection range of the LIDAR sensor by comparing the expected intensity at the location of the point to the intensity information for the point, wherein the effective detection range corresponds to a distance at which the LIDAR sensor detects an object and identifies one or more characteristics of the object according to one or more threshold minimums, wherein the one or more threshold minimums include a percentage recall on a Lambertian target; and
controlling a behavior of the autonomous vehicle based on the effective detection range.