Toyota Research Institute, Inc.

Peiyan Gong

Kun-Hsin Chen

Shunsho Kaku

Hai Jin

Sudeep Pillai

Arjun Bhargava

Ryan W Wolcott

Kuan-Hui Lee

Michael James Delp

Wolfram Burgard

Kuan-hui Lee

Christian Potthast

Yutaka Taruoka

Sarah Yoo

David L Garber

Systems, methods, computer-readable media, techniques, and methodologies are disclosed for performing end-to-end, learning-based keypoint detection and association. A scene graph of a signalized intersection is constructed from an input image of the intersection. The scene graph includes detected keypoints and linkages identified between the keypoints. The scene graph can be used along with a vehicle's localization information to identify which keypoint that represents a traffic signal is associated with the vehicle's current travel lane. An appropriate vehicle action may then be determined based on a transition state of the traffic signal keypoint and trajectory information for the vehicle. A control signal indicative of this vehicle action may then be output to cause an autonomous vehicle, for example, to implement the appropriate vehicle action.

End-to-end signalized intersection transition state estimator with scene graphs over semantic keypoints

A system and method are provided for making a probabilistic determination of the state of a lane transition which may be controlled by a traffic signal, such as a traffic cue. A determination of the state of transition of a traffic signal may be used by a vehicle to determine a course of action to take. In making the determination, the states of multiple traffic signals may be combined into one collection of elements that has values associated with the likelihood of the traffic signals being in a given state. Optionally, a determination may be made of whether a traffic signal is occluded.

Probabilistic modular lane transition state estimation

Systems and methods are provided for developing/updating training datasets for traffic light detection/perception models. V2I-based information may indicate a particular traffic light state/state of transition. This information can be compared to a traffic light perception prediction. When the prediction is inconsistent with the V2I-based information, data regarding the condition(s)/traffic light(s)/etc. can be saved and uploaded to a training database to update/refine the training dataset(s) maintained therein. In this way, an existing traffic light perception model can be updated/improved and/or a better traffic light perception model can be developed.

Traffic light active learning pipeline

Systems and methods for controlling the operation of an autonomous vehicle are disclosed herein. One embodiment performs traffic light detection at an intersection using a sensor-based traffic light detector to produce a sensor-based detection output, the sensor-based detection output having an associated first confidence level; performs traffic light detection at the intersection using a vehicle-to-infrastructure-based (V2I-based) traffic light detector to produce a V2I-based detection output, the V2I-based detection output having an associated second confidence level; performs one of (1) selecting as a final traffic-light-detection output whichever of the sensor-based detection output and the V2I-based detection output has a higher associated confidence level and (2) generating the final traffic-light-detection output by fusing the sensor-based detection output and the V2I-based detection output using a first learning-based classifier; and controls the operation of the autonomous vehicle based, at least in part, on the final traffic-light-detection output.

Systems and methods for controlling the operation of an autonomous vehicle using multiple traffic light detectors

A method for traffic light auto-labeling includes aggregating vehicle-to-infrastructure (V2I) traffic light signals at an intersection to determine transition states of each driving lane at the intersection during operation of an ego vehicle. The method also includes automatically labeling image training data to form auto-labeled image training data for a traffic light recognition model within the ego vehicle according to the determined transition states of each driving lane at the intersection. The method further includes planning a trajectory of the ego vehicle to comply with a right-of-way according to the determined transition states of each driving lane at the intersection according to a trained traffic light detection model. A federated learning module may train the traffic light recognition model using the auto-labeled image training data during the operation of the ego vehicle.

Traffic light detection auto-labeling and federated learning based on vehicle-to-infrastructure communications

Growing community of inventors

Ann Arbor, MI, United States of America

Peiyan Gong