Unsupervised velocity prediction and correction for urban driving entities from sequence of noisy position estimates

Patent metadata

Field	Value
Publication number	US-11724720-B2
Application number	US-202117487809-A
Country	US
Kind code	B2
Filing date	Sep 28, 2021
Priority date	Sep 28, 2021
Publication date	Aug 15, 2023
Grant date	Aug 15, 2023

Abstract

A method using unsupervised velocity prediction and correction for urban driving from sequences of noisy position estimates includes: performing a vehicle velocity prediction for one or more other vehicles in a vicinity of a host automobile vehicle; calculating a first heuristic based on a uniformity test; calculating a second heuristic based on a vehicle speed of the one or more other vehicles; combining the first heuristic and the second heuristic using a weighted sum; determining an uncertainty mask applying the combined first heuristic and the second heuristic and a heuristic threshold; and applying the uncertainty mask to identify a velocity correction for use by the host automobile vehicle.

First claim

What is claimed is:

1. A method for controlling a host vehicle comprising:
detecting a remote vehicle within perception data received from a sensor mounted to the host vehicle;
generating agent tokens from the perception data, wherein the agent tokens are associated with the remote vehicle at different times, and the agent tokens each include a time stamp, a remote vehicle identification, and a location of the remote vehicle;
predicting a velocity of the remote vehicle based on the agent tokens;
calculating heuristics for each of the agent tokens;
calculating a weighted sum of the heuristics;
comparing the weighted sum of the heuristics to a threshold to identity time stamps of the agent tokens where the velocity may have an error;
correcting the velocity of the agent tokens that have been identified as potentially having an error;
generating corrected agent tokens having the corrected velocity;
generating a motion plan based on the corrected agent tokens; and
controlling the host vehicle using steering, acceleration, and braking based on the motion plan.

2. The method of claim 1 , wherein calculating the heuristics includes:
calculating a first heuristic value based on a uniformity test;
calculating a second heuristic value based on a vehicle speed of the remote vehicle; and
combining the first heuristic and the second heuristic using a weighted sum.

3. The method of claim 1 , wherein correcting the velocity includes correcting a vehicle heading by interpolating between vehicle headings of closest confident neighbor agent tokens on each side of the agent token that has been identified as having a potential error.

4. The method of claim 1 , further including:
providing multiple agent tokens for one or more other vehicles in a vicinity of the host vehicle; and
individually incorporating a timestamp, a vehicle identification (ID), an “X” location for the one or more other vehicles, a “Y” location for the one or more other vehicles, a speed, and a heading for the multiple agent tokens.

5. A system for controlling a host vehicle, the system comprising:
a sensor mounted to the host vehicle;
a memory;
a processor in communication with the memory, wherein the processor executes instructions stored in the memory, which cause the processor to execute a method, the method comprising:
detecting a remote vehicle within perception data received from the sensor;
generating agent tokens from the perception data, wherein the agent tokens are associated with the remote vehicle at different times, and the agent tokens each include a time stamp, a remote vehicle identification, and a location of the remote vehicle;
predicting a velocity of the remote vehicle based on the agent tokens;
calculating heuristics for each of the agent tokens;
calculating a weighted sum of the heuristics;
comparing the weighted sum of the heuristics to a threshold to identity time stamps of the agent tokens where the velocity may have an error;
correcting the velocity of the agent tokens that have been identified as potentially having an error;
generating corrected agent tokens having the corrected velocity;
generating a motion plan based on the corrected agent tokens; and
controlling the host vehicle using steering, acceleration, and braking based on the motion plan.

6. A non-transitory computer readable medium having stored thereon instructions, which when executed by a processor cause the processor to execute a method, the method comprising:
detecting a remote vehicle within perception data received from a sensor mounted on a host vehicle;
generating agent tokens from the perception data, wherein the agent tokens are associated with the remote vehicle at different times, and the agent tokens each include a time stamp, a remote vehicle identification, and a location of the remote vehicle;
predicting a velocity of the remote vehicle based on the agent tokens;
calculating heuristics for each of the agent tokens;
calculating a weighted sum of the heuristics;
comparing the weighted sum of the heuristics to a threshold to identity time stamps of the agent tokens where the velocity may have an error;
correcting the velocity of the agent tokens that have been identified as potentially having an error;
generating corrected agent tokens having the corrected velocity;
generating a motion plan based on the corrected agent tokens; and
controlling the host vehicle using steering, acceleration, and braking based on the motion plan.

Assignees

• Gm Global Tech Operations Llc

Inventors

• Agarwal Amit
• Kwon Hyukseong
• Bhattacharyya Rajan
• Daily Michael J
• Holland Gavin D

Classifications

• B60W2556/50

  of positioning data, e.g. GPS [Global Positioning System] data · CPC title

• B60W60/0027 Primary

  using trajectory prediction for other traffic participants · CPC title

• B60W50/0098

  Details of control systems ensuring comfort, safety or stability not otherwise provided for · CPC title

• B60W2050/006

  Interpolation; Extrapolation · CPC title

• B60W2554/4042

  Longitudinal speed · CPC title