Modifying behavior of autonomous vehicles based on sensor blind spots and limitations
Models can be generated of a vehicle's view of its environment and used to maneuver the vehicle. This view need not include what objects or features the vehicle is actually seeing, but rather those areas that the vehicle is able to observe using its sensors if the sensors were completely un-occluded. For example, for each of a plurality of sensors of the object detection component, a computer may generate an individual 3D model of that sensor's field of view. Weather information is received and used to adjust one or more of the models. After this adjusting, the models may be aggregated into a comprehensive 3D model. The comprehensive model may be combined with detailed map information indicating the probability of detecting objects at different locations. The model of the vehicle's environment may be computed based on the combined comprehensive 3D model and detailed map information.
1. A method comprising:
generating, for each given sensor of a plurality of sensors for detecting objects in a vehicle's environment, a 3D model of the given sensor's field of view;
aggregating, by one or more processors, the plurality of 3D models to generate a comprehensive model, wherein the comprehensive model indicates an extent of an aggregated field of view for the plurality of sensors;
combining the comprehensive model with map information corresponding to environmental data for the vehicle's environment obtained at a previous point in time using probability data of the map information indicating a probability of detecting objects at various locations in the map information from various possible locations of the vehicle to produce a combined model annotated with information identifying a first portion of the environment as occupied by an object, a second portion of the environment as unoccupied by an object, and a third portion of the environment as unobserved by any of the plurality of sensors; and
using the combined model to maneuver the vehicle.
2. The method of claim 1 , wherein the 3D model of each given sensor's field of view corresponds to a pre-determined model of the given sensor's unobstructed field of view.
3. The method of claim 1 , wherein the 3D model for each given sensor's field of view is based on the given sensor's location and orientation relative to the vehicle.
4. The method of claim 1 , further comprising, prior to combining, adjusting one or more characteristics of the plurality of 3D models by building at least one parameterized model for a set of weather conditions.
5. The method of claim 1 , further comprising, prior to combining, adjusting one or more characteristics of the plurality of 3D models by reducing the effective range for a given sensor.
6. The method of claim 1 , wherein the probability of a probability of detecting objects at various locations in the map information includes a probability of detecting another object of a particular size not already included in the map information on an opposite side of a given object defined in the map information.
7. The method of claim 1 , wherein the probability of a probability of detecting objects at various locations in the map information includes a probability of detecting another object of a particular shape not already included in the map information on an opposite side of a given object defined in the map information.
8. The method of claim 1 , further comprising:
receiving information about a current state of weather in the vehicle's environment; and
prior to aggregating, adjusting a shape of at least one model of the plurality of 3D models using the current state of weather in real time.
9. The method of claim 1 , wherein the comprehensive model includes a binary map indicating areas of vehicle's environment that the plurality of sensors is currently able to observe.
10. The method of claim 1 , wherein using the combined model to maneuver the vehicle includes slowing the vehicle down.
11. The method of claim 1 , wherein using the combined model to maneuver the vehicle includes repositioning the vehicle to improve the comprehensive model.
12. The method of claim 1 , wherein using the combined model to maneuver the vehicle includes avoiding certain maneuvers.
13. A system comprising:
one or more processors configured to:
generate, for each given sensor of a plurality of sensors for detecting objects in a vehicle's environment, a 3D model of the given sensor's field of view;
aggregate, by one or more processors, the plurality of 3D models to generate a comprehensive model, wherein the comprehensive model indicates an extent of an aggregated field of view for the plurality of sensors;
combine the comprehensive model with map information corresponding to environmental data for the vehicle's environment obtained at a previous point in time using probability data of the map information indicating a probability of detecting objects at various locations in the map information from various possible locations of the vehicle to produce a combined model annotated with information identifying a first portion of the environment as occupied by an object, a second portion of the environment as unoccupied by an object, and a third portion of the environment as unobserved by any of the plurality of sensors; and
use the combined model to maneuver the vehicle.
14. The system of claim 13 , wherein the one or more processors are further configured to, prior to combining, adjust one or more characteristics of the plurality of 3D models by building at least one parameterized model for a set of weather conditions.
15. The system of claim 13 further comprising the plurality of sensors and the vehicle.
16. The system of claim 13 , wherein the probability of a probability of detecting objects at various locations in the map information includes a probability of detecting another object of a particular size not already included in the map information on an opposite side of a given object defined in the map information.
17. The system of claim 13 , wherein the probability of a probability of detecting objects at various locations in the map information includes a probability of detecting another object of a particular shape not already included in the map information on an opposite side of a given object defined in the map information.
18. The system of claim 13 , wherein the one or more processors are further configured to:
receive information about a current state of weather in the vehicle's environment; and
prior to aggregating, adjust a shape of at least one model of the plurality of 3D models using the current state of weather in real time.
19. The system of claim 13 , wherein the comprehensive model includes a binary map indicating areas of vehicle's environment that the plurality of sensors is currently able to observe.
20. A non-transitory, computer readable recording medium on which instructions are stored, the instructions, when executed by one or more processors cause the one or more processors to perform a method, the method comprising:
generating, for each given sensor of a plurality of sensors for detecting objects in a vehicle's environment, a 3D model of the given sensor's field of view;
aggregating the plurality of 3D models to generate a comprehensive model, wherein the comprehensive model indicates an extent of an aggregated field of view for the plurality of sensors;
combining the comprehensive model with map information corresponding to environmental data for the vehicle's environment obtained at a previous point in time using probability data of the map information indicating a probability of detecting objects at various locations in the map information from various possible locations of the vehicle to produce a combined model annotated with information identifying a first portion of the environment as occupied by an object, a second portion of the environment as unoccupied by an object, and a third portion of the environment as unobserved by any of the plurality of sensors; and
using the combined model to maneuver the vehicle.