Apollo Autonomous Driving USA LLC

Baidu USA LLC

Beijing Baidu Netcom Science Technology Co., Ltd

Szu-Hao Wu

Shu Jiang

Hao Liu

Helen K Pan

Yu Long Cao

Jiangtao Hu

Weiman Lin

Haofeng Kou

Xiaoyi Zhu

Jeong Ho Lyu

Ang Li

Linpeng Cheng

Fuyang Zhao

The present disclosure provides methods and techniques for evaluating and improving algorithms for autonomous driving planning and control (PNC), using one or more metrics (e.g., similarity scores) computed based on expert demonstrations. For example, the one or more metrics allow for improving PNC based on human, as opposed to or in addition to optimizing certain oversimplified properties, such as the least distance or time, as an objective. When driving in certain scenarios, such as taking a turn, people may drive in a distributed probability pattern instead of in a uniform line (e.g., different speeds and different curvatures at the same corner). As such, there can be more than one 'correct' control trajectory for an autonomous vehicle to perform in the same turn. Safety, comfort, speeds, and other criteria may lead to different preferences and judgment as to how well the controlled trajectory has been computed.

Distributional expert demonstrations for autonomous driving

In one embodiment, a system configures an image signal processor (ISP) of an autonomous driving vehicle (ADV) with a first set of ISP configuration parameters, where the ISP is used to process raw image data of an image sensor of the ADV based on the first set of ISP configuration parameters. The system determines whether one or more criteria is satisfied, where the one or more criteria corresponds to an expected change in a characteristic of ambient light being perceived by the image sensor of the ADV. In response to determining that the one or more criteria is satisfied, the system configures the ISP of the ADV with a second set of ISP configuration parameters, where the ISP is used to apply an image processing algorithm to raw image data based on the second set of ISP configuration parameters to generate an image.

On-board tuning of image signal processor for cameras of autonomous vehicles

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.

On-board parameter tuning for control module for autonomous vehicles

Growing community of inventors

Sunnyvale, CA, United States of America

Szu-Hao Wu