Back to results
US 12,576,836 B2Granted
Conditional object position prediction by a machine learned model
AssigneeZoox, Inc.
InventorsGowtham Garimella, Carl-Johan E Hoel, Xianan Huang, Marin Kobilarov, Gregory Michael Woelki, Albert Zhao
Life of this patent
Estimating the term…
Abstract
Techniques for predicting positions of objects in an environment are described herein. For example, the techniques may include clustering predicted object positions and using a representative target for one or more clusters to determine a predicted position of an autonomous vehicle. The clusters may vary in size, number of positions, etc. Object positions can be selected from one or more clusters for consideration during vehicle planning, which may include simulation.
Claims20
A system comprising:
- one or more processors;
- and
- one or more non-transitory computer-readable media storing instructions executable by the one or more processors, wherein the instructions, when executed, cause the system to perform actions comprising:
- receiving, by a first machine learned model, first data indicating historical locations associated with multiple objects in an environment;
- predicting, by the first machine learned model and based at least in part on the first data, a first set of candidate positions associated with an object;
- receiving, by a second machine learned model, second data indicating locations associated with an autonomous vehicle in the environment, the second data different from the first data;
- predicting, by the second machine learned model and a combination of the first data and the second data, a second set of candidate positions associated with the autonomous vehicle;
- clustering the first set of candidate positions associated with the object relative to the second set of candidate positions associated with the autonomous vehicle;
- determining, based at least in part on the clustering, third data representing a subset of the first set of candidate positions associated with the object;
- and
- transmitting the third data to a vehicle computing device associated with the autonomous vehicle, the vehicle computing device configured to determine an action for the autonomous vehicle to navigate in relation to the object.
8 more elements
A method comprising:
- determining candidate object positions for an object based at least in part on historical data associated with multiple objects in an environment, the multiple objects comprising the object;
- determining a first candidate vehicle position and a second candidate vehicle position for a vehicle to occupy at a future time;
- clustering, into a first cluster, a first portion of the candidate object positions relative to the first candidate vehicle position for the vehicle;
- clustering, into a second cluster a second portion of the candidate object positions relative to the second candidate vehicle position for the vehicle;
- determining a subset of object positions from one or more of:
- the first portion of the candidate object positions of the first cluster or the second portion of the candidate object positions of the second cluster;
- and
- transmitting the subset of object positions to a vehicle computing device for determining an action for the vehicle to navigate in relation to the object.
5 more elements
One or more non-transitory computer-readable media storing instructions that, when executed, cause one or more processors to perform actions comprising:
- determining candidate object positions for an object based at least in part on historical data associated with multiple objects in an environment, the multiple objects comprising the object;
- determining a first candidate vehicle position and a second candidate vehicle position for a vehicle to occupy at a future time;
- clustering, into a first cluster, a first portion of the candidate object positions relative to the first candidate vehicle position for the vehicle;
- clustering, into a second cluster, a second portion of the candidate object positions relative to the second candidate vehicle position for the vehicle;
- determining a subset of object positions from one or more of:
- the first portion of the candidate object positions of the first cluster or the second portion of the candidate object positions of the second cluster;
- and
- transmitting the subset of object positions to a vehicle computing device for determining an action for the vehicle to navigate in relation to the object.
5 more elements
Dependencies are inferred from the claim text.