Collision aware path planning systems and methods
Systems and methods for path planning for autonomous vehicles err on the side of a safer road position and safer type of collision if one were to occur. An autonomous vehicle perception module detects target objects within a region of interest (ROI). An autonomous vehicle processor estimates relative velocity between the ego vehicle and target objects, and estimates mass of target objects. Collision-aware path planning calculates a cost function that assigns higher costs for paths that take the vehicle close to objects that have a large velocity difference and higher costs for objects with large estimated mass. Path planning provides a cost map that yields a path that has appropriate buffer distances between the autonomous vehicle and surrounding objects. In the event the ego vehicle balances buffer distance margins to multiple surrounding target objects, target objects that would create more severe collisions are given greater buffer distance.
1 . A method comprising:
receiving, by a processor associated with an autonomous vehicle, an indication that a current trajectory of the autonomous vehicle is associated with a likelihood of a collision that satisfies a threshold indicating a potential collision;
identifying, by the processor, an alternative trajectory for the vehicle;
when the alternative trajectory has a likelihood of collision that satisfies the threshold indicating a potential collision, calculating, by the processor, a cost value for the current trajectory and the alternative trajectory, wherein the cost value is based on a speed of a second vehicle associated with the potential collision; and
transmitting, by the processor, an input of a trajectory having a lowest cost value to a vehicle control module of the autonomous vehicle.
2 . The method of claim 1 , wherein the cost value is further based on an attribute of an additional object associated with the potential collision.
3 . The method of claim 2 , wherein the attribute is at least one of a mass, volume, density, or material of the object.
4 . The method of claim 2 , further comprising:
executing, by the processor, an artificial intelligence model to predict the attribute of the object, the artificial intelligence model configured to ingest data from at least one sensor of the autonomous vehicle and predict the attribute of the object.
5 . A method comprising:
identifying, by a processor associated with an autonomous vehicle, a plurality of target objects in collision proximity to the autonomous vehicle;
executing, by the processor, an artificial intelligence model to predict a plurality of predetermined attributes of each of the plurality of target objects relative to the autonomous vehicle;
calculating, by the processor, a total cost value representing a total of respective cost values incurred from each of the plurality of target objects based on the plurality of predicted attributes;
determining, by the processor, a plurality of vehicle poses of the autonomous vehicle; and
transmitting to a vehicle control module of the autonomous vehicle, by the processor, an input of a vehicle pose selected from the plurality of vehicle poses having a lowest total cost value.
6 . The method of claim 5 , wherein each vehicle pose of the plurality of vehicle poses has a respective set of buffer distance margins between the autonomous vehicle and respective target objects of the plurality of target objects.
7 . The method of claim 5 , further comprising repeating the executing step and calculating step for each vehicle pose of the plurality of vehicle poses.
8 . The method of claim 7 , wherein the lowest total cost value is a lowest total cost value of the calculated total cost values.
9 . The method of claim 5 , wherein the total cost value representing the total of respective cost values incurred from each of the plurality of target objects represents a collision severity.
10 . The method of claim 5 , wherein the plurality of predetermined attributes of each of the plurality of target objects comprise a relative velocity of a respective target object of the plurality of target objects relative to the autonomous vehicle and a mass attribute of the respective target object.
11 . The method of claim 5 , wherein the artificial intelligence model is configured to ingest data from at least one sensor of the autonomous vehicle and predict the plurality of predetermined attributes of each of the target objects.
12 . The method of claim 5 , wherein identifying the plurality of target objects in collision proximity to the autonomous vehicle comprises identifying objects within a region of interest (ROI) that satisfy predetermined criteria for likelihood of collision with the autonomous vehicle.
13 . A system comprising:
a non-transitory computer-readable medium comprising instructions that are configured to be executed by at least one processor associated with an automated vehicle to:
receive an indication that a current trajectory of the autonomous vehicle is associated with a likelihood of a collision that satisfies a threshold indicating a potential collision;
identify an alternative trajectory for the vehicle;
when the alternative trajectory has a likelihood of collision that satisfies the threshold indicating a potential collision, calculate a cost value for the current trajectory and the alternative trajectory, wherein the cost value is based on a speed of a second vehicle associated with the potential collision; and
transmit an input of a trajectory having a lowest cost value to a vehicle control module of the autonomous vehicle.
14 . The system of claim 13 , wherein the cost value is further based on an attribute of an additional object associated with the potential collision.
15 . The system of claim 14 , wherein the attribute is at least one of a mass, volume, density, or material of the object.
16 . The system of claim 14 , wherein the at least one processor is further configured to execute an artificial intelligence model to predict the attribute of the object, the artificial intelligence model configured to ingest data from at least one sensor of the autonomous vehicle and predict the attribute of the object.