On-board parameter tuning for control module for autonomous vehicles

Patent metadata

Field	Value
Publication number	US-12195019-B2
Application number	US-202218003017-A
Country	US
Kind code	B2
Filing date	Nov 28, 2022
Priority date	Nov 28, 2022
Publication date	Jan 14, 2025
Grant date	Jan 14, 2025

Abstract

In one embodiment, a microcontroller unit (MCU) receives an expected state of an autonomous driving vehicle (ADV) from a controller of the ADV, where the controller controls motions of the ADV using a control algorithm. The MCU receives sensor data from one or more sensors of the ADV. The MCU determine an actual state of the ADV based on the sensor data. The MCU determines a performance metric of the control algorithm based on the expected state and the actual state. In response to determining the performance metric has satisfied a predetermined condition, the MCU determines a plurality of weight values for the control algorithm. The MCU sends the plurality of weight values to the control system to tune one or more weight parameters of the control algorithm using the plurality of weight values, where the controller controls the ADV using the tuned control algorithm.

First claim

What is claimed is:

1. A computer-implemented method for control parameter tuning, comprising:
receiving, at a microcontroller unit (MCU), an expected state of an autonomous driving vehicle (ADV) from a controller of the ADV, wherein the controller controls motions of the ADV using a control algorithm;
receiving, at the MCU, sensor data from one or more sensors of the ADV;
determining, at the MCU, an actual state of the ADV based on the sensor data;
determining, at the MCU, a performance metric of the control algorithm based on the expected state and the actual state;
in response to determining the performance metric has satisfied a predetermined condition, determining a plurality of weight values for the control algorithm; and
sending the plurality of weight values to the controller to tune one or more weight parameters of the control algorithm using the plurality of weight values, wherein the controller controls the ADV using the tuned control algorithm.

2. The method of claim 1 , further comprising:
determining a tracking error based on a difference between the expected state and the actual state; and
determining the performance metric based on the tracking error.

3. The method of claim 1 , wherein determining a plurality of weight values for the control algorithm comprises: selecting the plurality of weight values based on a rule-based algorithm, wherein the rule-based algorithm uses historical states of the ADV, control inputs, and/or one or more environment variables as inputs to select the plurality of weight values.

4. The method of claim 1 , wherein determining the plurality of weight values for the control algorithm comprises: generating the plurality of weight values based on a machine learning algorithm, wherein the machine learning algorithm is trained using historical data including historical states of the ADV, control inputs, one or more environment variables, and one or more parameters of the control algorithm.

5. The method of claim 1 , wherein the performance metric has satisfied a predetermined condition if the performance metric has degraded from a previous iteration, wherein the performance metric has degraded if a tracking error of the control algorithm has increased.

6. The method of claim 5 , wherein the performance metric is measured at a dynamic frequency, wherein when control performance degrades, the performance metric is measured at a first frequency and when the control performance improves, the performance metric is measured at a second frequency, wherein the second frequency is less than the first frequency.

7. The method of claim 1 , wherein the expected state or the actual state includes a velocity, an acceleration, or a location of the ADV.

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, at a microcontroller unit (MCU), an expected state of an autonomous driving vehicle (ADV) from a controller of the ADV, wherein the controller controls motions of the ADV using a control algorithm;
receiving, at the MCU, sensor data from one or more sensors of the ADV;
determining, at the MCU, an actual state of the ADV based on the sensor data;
determining, at the MCU, a performance metric of the control algorithm based on the expected state and the actual state;
in response to determining the performance metric has satisfied a predetermined condition, determining a plurality of weight values for the control algorithm; and
sending the plurality of weight values to the controller to tune one or more weight parameters of the control algorithm using the plurality of weight values, wherein the controller controls the ADV using the tuned control algorithm.

9. The non-transitory machine-readable medium of claim 8 , wherein the operations further comprise:
determining a tracking error based on a difference between the expected state and the actual state; and
determining the performance metric based on the tracking error.

10. The non-transitory machine-readable medium of claim 8 , wherein determining a plurality of weight values for the control algorithm comprises: selecting the plurality of weight values based on a rule-based algorithm, wherein the rule-based algorithm uses historical states of the ADV, control inputs, and/or one or more environment variables as inputs to select the plurality of weight values.

11. The non-transitory machine-readable medium of claim 8 , wherein determining the plurality of weight values for the control algorithm comprises: generating the plurality of weight values based on a machine learning algorithm, wherein the machine learning algorithm is trained using historical data including historical states of the ADV, control inputs, one or more environment variables, and one or more parameters of the control algorithm.

12. The non-transitory machine-readable medium of claim 8 , wherein the performance metric has satisfied a predetermined condition if the performance metric has degraded from a previous iteration, wherein the performance metric has degraded if a tracking error of the control algorithm has increased.

13. The non-transitory machine-readable medium of claim 12 , wherein the performance metric is measured at a dynamic frequency, wherein when control performance degrades, the performance metric is measured at a first frequency and when the control performance improves, the performance metric is measured at a second frequency, wherein the second frequency is less than the first frequency.

14. The non-transitory machine-readable medium of claim 8 , wherein the expected state or the actual state includes a velocity, an acceleration, or a location of the ADV.

15. A microcontroller unit, 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, an expected state of an autonomous driving vehicle (ADV) from a controller of the ADV, wherein the controller controls motions of the ADV using a control algorithm;
receiving, sensor data from one or more sensors of the ADV;
determining, an actual state of the ADV based on the sensor data;
determining, a performance metric of the control algorithm based on the expected state and the actual state;
in response to determining the performance metric has satisfied a predetermined condition, determining a plurality of weight values for the control algorithm; and
sending the plurality of weight values to the controller to tune one or more weight parameters of the control algorithm using the plurality of weight values, wherein the controller controls the ADV using the tuned control algorithm.

16. The microcontroller unit of claim 15 , wherein the operations further comprise:
determining a tracking error based on a difference between the expected state and the actual state; and
determining the performance metric based on the tracking error.

17. The microcontroller unit of claim 15 , wherein determining a plurality of weight values for the control algorithm comprises: selecting the plurality of weight values based on a rule-based algorithm, wherein the rule-based algorithm uses historical states of the ADV, control inputs, and/or one or more environment variables as inputs to select the plurality of weight values.

18. The microcontroller unit of claim 15 , wherein determining the plurality of weight values for the control algorithm comprises: generating the plurali

Assignees

• Apollo Autonomous Driving USA LLC
• Beijing Baidu Netcom Sci & Tech Co Ltd

Inventors

• Jiang Shu
• Liu Hao
• Wu Szu-Hao
• Zhao Fuyang
• Zhu Xiaoyi
• Kou Haofeng
• Pan Helen K

Classifications

• B60W2050/0022

  Gains, weighting coefficients or weighting functions · CPC title

• B60W2520/105

  Longitudinal acceleration · CPC title

• B60W2520/10

  Longitudinal speed · CPC title

• B60W2556/10

  Historical data · CPC title

• B60W60/00

  Drive control systems specially adapted for autonomous road vehicles · CPC title