Autonomous vehicle routing using annotated maps
A control system for an autonomous vehicle can determine a risk value for each respective path segment of a plurality of path segments in a given area that includes a destination of the autonomous vehicle. The risk value can correspond to a cost layer in a map that includes the respective path segment. Based on the risk value for each respective path segment, the control system can determine a travel route for the autonomous vehicle to the destination, and autonomously control the autonomous vehicle to navigate along the travel route to the destination.
1. A control system for an autonomous vehicle, comprising:
one or more processors; and
one or more memory resources storing instructions that, when executed by the one or more processors, cause the control system to:
determine a risk value for each respective path segment of a plurality of path segments in a given area that includes a destination of the autonomous vehicle, the risk value corresponding to a cost layer in a map that includes the respective path segment;
based on the risk value for each respective path segment, determine a travel route for the autonomous vehicle to the destination; and
autonomously control the autonomous vehicle to navigate along the travel route to the destination.
2. The control system of claim 1 , wherein the executed instructions cause the control system to determine the risk value for each respective path segment of the plurality of path segments using a weighted sum of (i) scores for each of a plurality of risk factors for the respective path segment, and (ii) a travel time calculation for the respective path segment.
3. The control system of claim 2 , wherein the scores for each of the plurality of risk factors represents probabilities that an undesirable event may occur during autonomous operation of the autonomous vehicle along the respective path segment.
4. The control system of claim 2 , wherein the scores for each of the plurality of risk factors and the travel time calculation are determined through execution of a statistical model.
5. The control system of claim 4 , wherein the statistical model determines the scores for each of the plurality of risk factors based, at least in part, on sensor data collected from a plurality of sensors coupled to the autonomous vehicle and from other vehicles operating throughout the given area.
6. The control system of claim 5 , wherein the executed instructions further cause the control system to:
analyze the sensor data to identify events that impact one or more of the plurality of risk factors.
7. The control system of claim 5 , wherein the executed instructions further cause the control system to:
determine, from the sensor data, one or more road conditions for a current path segment that affect one or more of the plurality of risk factors; and
autonomously control the autonomous vehicle to navigate along an alternate path segment based on the determined risk values for the plurality of path segments.
8. The control system of claim 1 , wherein the determined risk value for each respective path segment of the plurality of path segments represents a cost of autonomous operation of the autonomous vehicle along the respective path segment, and wherein the executed instructions cause the control system to generate the travel route based on a sum of the determined risk value for each of the plurality of path segments.
9. The control system of claim 1 , wherein the plurality of path segments correspond to lanes of roads identified in a set of localization maps for the given area.
10. The control system of claim 2 , wherein the plurality of risk factors comprise (i) an intervention risk corresponding to when the autonomous vehicle is incapable of autonomous operation along a given path segment, (ii) a bad experience risk corresponding to hard stops, jerking motions, and other events that impact a passenger experience, and (iii) a harmful event risk corresponding to a collision involving the autonomous vehicle.
11. A method of controlling navigation of an autonomous vehicle, the method being implemented by one or more processors and comprising:
determining a risk value for each respective path segment of a plurality of path segments in a given area that includes a destination of the autonomous vehicle, the risk value corresponding to a cost layer in a map that includes the respective path segment;
based on the risk value for each respective path segment, determining a travel route for the autonomous vehicle to the destination; and
autonomously controlling the autonomous vehicle to navigate along the travel route to the destination.
12. The method of claim 11 , wherein the one or more processors determine the risk value for each respective path segment of the plurality of path segments using a weighted sum of (i) scores for each of a plurality of risk factors for the respective path segment, and (ii) a travel time calculation for the respective path segment.
13. The method of claim 12 , wherein the scores for each of the plurality of risk factors represents probabilities that an undesirable event may occur during autonomous operation of the autonomous vehicle along the respective path segment.
14. The method of claim 12 , wherein the scores for each of the plurality of risk factors and the travel time calculation are determined through execution of a statistical model.
15. The method of claim 14 , wherein the statistical model determines the scores for each of the plurality of risk factors based, at least in part, on sensor data collected from a plurality of sensors coupled to the autonomous vehicle and from other vehicles operating throughout the given area.
16. The method of claim 15 , wherein further comprising:
analyzing the sensor data to identify events that impact one or more of the plurality of risk factors.
17. The method of claim 15 , wherein further comprising:
determining, from the sensor data, one or more road conditions for a current path segment that affect one or more of the plurality of risk factors; and
autonomously controlling the autonomous vehicle to navigate along an alternate path segment based on the determined risk values for the plurality of path segments.
18. The method of claim 11 , wherein the determined risk value for each respective path segment of the plurality of path segments represents a cost of autonomous operation of the autonomous vehicle along the respective path segment, and wherein the one or more processors generate the travel route based on a sum of the determined risk value for each of the plurality of path segments.
19. The method of claim 11 , wherein the plurality of path segments correspond to lanes of roads identified in a set of localization maps for the given area.
20. An autonomous vehicle comprising:
a plurality of sensors that generate sensor data corresponding to autonomous operation of the autonomous vehicle;
one or more processors; and
one or more memory resources storing instructions that, when executed by the one or more processors, cause the one or more processors to:
determine a risk value for each respective path segment of a plurality of path segments in a given area that includes a destination of the autonomous vehicle, the risk value corresponding to a cost layer in a map that includes the respective path segment;
based on the risk value for each respective path segment, determine a travel route for the autonomous vehicle to the destination; and
autonomously control the autonomous vehicle to navigate along the travel route to the destination.