Labeling scheme for labeling and generating high-definition map based on trajectories driven by vehicles
According to one embodiment, perception data describing a set of trajectories driven by a number of vehicles is received at a server from the vehicles or data collection agents over a network. The vehicles were driving through a road segment of a road over a period of time and their driving trajectories were captured. A trajectory analysis is the performed on the perception data using a set of rules to determine driving behaviors of the corresponding vehicles. A lane configuration of the road segment is then determined based on the driving behaviors. A map segment of a navigation map is then updated based on the lane configuration of one or more lanes within the road segment. A higher definition map can be generated based on the updates of the navigation map and the lane configuration, which can be utilized to autonomously drive an ADV subsequently.
1. A computer-implemented method for generating a map for autonomous driving, the method comprising:
receiving perception data describing a set of trajectories driven by a plurality of vehicles navigating through a road segment of a road over a period of time;
performing an analysis using a set of rules on the set of trajectories to determine driving behaviors of the plurality of vehicles;
determining a lane configuration of one or more lanes of the road segment based on the driving behaviors of the plurality of vehicles, comprising:
determining a number of lanes within the road segment based on the driving behaviors of the plurality of vehicles within the road segment; and
for each of the lanes, determining a lane width based on trajectory patterns of a plurality of trajectories within the lane;
labelling a map segment of a navigation map corresponding to the road segment to include lane configuration information of the one or more lanes to generate a higher definition (HD) map, wherein the lane configuration information includes the number of lanes and the lane width of each lane, wherein the HD map is utilized to generate a path to autonomously drive an autonomous driving vehicle subsequently; and
transferring the HD map to the autonomous driving vehicle to enable controlling the autonomous driving vehicle to travel on the road segment according to the path generated based on the HD map.
2. The method of claim 1 , further comprising measuring a lane boundary of each lane based on the plurality of trajectories, wherein the lane configuration information includes the lane boundary of each lane.
3. The method of claim 1 , wherein determining the lane configuration of one or more lanes of the road segment comprises:
identifying an area that the plurality of vehicles drove to avoid based on the driving behaviors; and determining that a static obstacle is located within the avoided area based on the set of rules.
4. The method of claim 3 , further comprising measuring a size of the avoided area based on the plurality of trajectories near the avoided area, wherein the size of the avoided area is included in the HD map.
5. The method of claim 1 , wherein determining the lane configuration of one or more lanes of the road segment comprises detecting an intersection based on the plurality of trajectories of the plurality of vehicles.
6. The method of claim 1 , wherein determining the lane configuration of one or more lanes of the road segment comprises detecting a stop sign or a stop line based on the plurality of trajectories of the plurality of vehicles.
7. The method of claim 1 , wherein determining the lane configuration of one or more lanes of the road segment comprises detecting a traffic light based on the perception data.
8. A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations, the operations comprising:
receiving perception data describing a set of trajectories driven by a plurality of vehicles navigating through a road segment of a road over a period of time;
performing an analysis using a set of rules on the set of trajectories to determine driving behaviors of the plurality of vehicles;
determining a lane configuration of one or more lanes of the road segment based on the driving behaviors of the plurality of vehicles, comprising:
determining a number of lanes within the road segment based on the driving behaviors of the plurality of vehicles within the road segment; and
for each of the lanes, determining a lane width based on trajectory patterns of a plurality of trajectories within the lane;
labelling a map segment of a navigation map corresponding to the road segment to include lane configuration information of the one or more lanes to generate a higher definition (HD) map, wherein the lane configuration information inlcudes the number of lanes and the lane width of each lane, wherein the HD map is utilized to generate a path to autonomously drive an autonomous driving vehicle subsequently; and
transferring the HD map to the autonomous driving vehicle to enable controlling the autonomous driving vehicle to travel on the road segment according to the path generated based on the HD map.
9. The machine-readable medium of claim 8 , wherein the operation further comprise measuring a lane boundary of each lane based on the plurality of trajectories, wherein the lane configuration information includes the lane boundary of each lane.
10. The machine-readable medium of claim 8 , wherein determining the lane configuration of one or more lanes of the road segment comprises:
identifying an area that the plurality of vehicles drove to avoid based on the driving behaviors; and
determining that a static obstacle is located within the avoided area based on the set of rules.
11. The machine-readable medium of claim 10 , wherein the operations further comprise measuring a size of the avoided area based on the plurality of trajectories near the avoided area, wherein the size of the avoided area is included in the HD map.
12. The machine-readable medium of claim 8 , wherein determining the lane configuration of one or more lanes of the road segment comprises detecting an intersection based on the plurality of trajectories of the plurality of vehicles.
13. The machine-readable medium of claim 8 , wherein determining the lane configuration of one or more lanes of the road segment comprises detecting a stop sign or a stop line based on the plurality of trajectories of the plurality of vehicles.
14. The machine-readable medium of claim 8 , wherein determining the lane configuration of one or more lanes of the road segment comprises detecting a traffic light based on the perception data.
15. A data processing system, comprising:
a processor; and
a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations, the operations including:
receiving perception data describing a set of trajectories driven by a plurality of vehicles navigating through a road segment of a road over a period of time,
performing an analysis using a set of rules on the set of trajectories to determine driving behaviors of the plurality of vehicles,
determining a lane configuration of one or more lanes of the road segment based on the driving behaviors of the plurality of vehicles, comprising:
determining a number of lanes within the road segment based on the driving behaviors of the plurality of vehicles within the road segment; and
for each of the lanes, determining a lane width based on trajectory patterns of a plurality of trajectories within the lane,
updating a map segment of a navigation map corresponding to the road segment to include lane configuration information of the one or more lanes togenerate a higher definition (HD) map, wherein the lane configuration information includes the number of lanes and the lane width of each lane, wherein the HD map is utilized to generate a path to autonomously drive an autonomous driving vehicle subsequently; and
transferring the HD map to the autonomous driving vehicle to enable controlling the autonomous driving vehicle to travel on the road segment according to the path generated based on the HD map.
16. The system of claim 15 , wherein the operation further comprise measuring a lane boundary of each lane based on the plurality of trajectories, wherein the lane configuration information includes the lane boundary of each lane.
17. The system of claim 15 , wherein determining the lane configuration of one or more lanes of the road segment comprises:
identifying an area that the plurality of vehicles drove to avoid based on the driving behaviors; and
determining that a static obstacle is located within the avoided area based on the set of rules.