公开(公告)号：US20240010196A1

公开(公告)日：2024-01-11

申请号：US18183566

申请日：2023-03-14

Applicant: NVIDIA Corporation

Inventor： Karen Yan Ming Leung ,  Sushant Veer ,  Edward Fu Schmerling ,  Marco Pavone

IPC: B60W30/095 ,  B60W60/00 ,  B60W30/09

CPC classification number: B60W30/0956 ,  B60W60/0015 ,  B60W30/09 ,  B60W2554/4041

Abstract: In various examples, control policies for controlling agents may be learned from demonstrations capturing joint states of entities navigating through the environment. A control policy may be learned mapping joint states to control actions, where the joint states are between agents, and the control actions are of at least one of the agents. The control policy may be learned to define the mappings as control invariant sets of the joint sates and the control actions. The control policy may be used to determine one or more functions that compute, based at least on a joint state between entities, output indicating a likelihood of collision between the entities operating in accordance with the control policy. Using the output, current and/or potential states of the environment may be evaluated to determine control operations for a machine, such as a vehicle.

公开(公告)号：US20250029489A1

公开(公告)日：2025-01-23

申请号：US18379599

申请日：2023-10-12

Applicant: NVIDIA Corporation

Inventor： Yulong Cao ,  Chaowei Xiao ,  Marco Pavone ,  Boris Ivanovic

IPC: G08G1/0967 ,  G06N3/02 ,  G08G1/01

Abstract: In various examples, a traffic model including one or more traffic scenarios may be generated and/or updated based on using human feedback. Human feedback may be provided indicating a preference for various traffic scenarios to identify which scenarios in a model are more realistic. A reward model may capture the preference information and rank the realism of one or more traffic scenarios.

公开(公告)号：US20240199074A1

公开(公告)日：2024-06-20

申请号：US18335028

申请日：2023-06-14

Applicant: NVIDIA Corporation

Inventor： Sushant Veer ,  Karen Leung ,  Ryan Cosner ,  Yuxiao Chen ,  Marco Pavone

IPC: B60W60/00 ,  B60W40/02

CPC classification number: B60W60/0011 ,  B60W40/02

Abstract: Autonomous vehicles (AVs) may need to contend with conflicting traveling rules and the AV controller would need to select the least objectionable control action. A rank-preserving reward function can be applied to trajectories derived from a rule hierarchy. The reward function can be correlated to a robustness vector derived for each trajectory. Thereby the highest ranked rules would result in the highest reward, and the lower ranked rules would result in lower reward. In some aspects, one or more optimizers, such as a stochastic optimizer can be utilized to improve the results of the reward calculation. In some aspects, a sigmoid function can be applied to the calculation to smooth out the step function used to calculate the robustness vector. The preferred trajectory selected using the results from the reward function can be communicated to an AV controller for implementation as a control action.

公开(公告)号：US20240199068A1

公开(公告)日：2024-06-20

申请号：US18244050

申请日：2023-09-08

Applicant: NVIDIA Corporation

Inventor： Marco Pavone ,  Heng Yang

IPC: B60W60/00 ,  B60W50/00 ,  G06T7/73

CPC classification number: B60W60/001 ,  B60W50/0097 ,  G06T7/73 ,  B60W2420/42 ,  G06T2200/24 ,  G06T2207/20081 ,  G06T2207/20084 ,  G06T2207/20092 ,  G06T2207/30252

Abstract: Apparatuses, systems, and techniques to obtain prediction set(s) (e.g., region(s)) for keypoint prediction(s) based at least in part on data associated with an object, compute a set of candidate poses for the object based at least in part on the prediction set(s), and estimate an estimated object pose based at least in part on the set of candidate poses. The estimated object pose may be used to move a device. For example the estimated object pose may be used to provide collision-free motion generation for a real-world or virtual device (e.g., a robot, an autonomous machine, or a semi-autonomous machine). In at least one embodiment, at least a portion of the object pose estimation and/or at least a portion of the collision-free motion generation is performed in parallel.

公开(公告)号：US20240010232A1

公开(公告)日：2024-01-11

申请号：US18318233

申请日：2023-05-16

Applicant: NVIDIA Corporation

Inventor： Peter Karkus ,  Boris Ivanovic ,  Shie Mannor ,  Marco Pavone

IPC: B60W60/00 ,  B60W30/095 ,  B60W50/00 ,  G06N3/08

CPC classification number: B60W60/0011 ,  B60W30/0956 ,  B60W50/0097 ,  G06N3/08 ,  B60W2554/4041 ,  B60W2554/4045

Abstract: In various examples, a motion planner include an analytical function to predict motion plans for a machine based on predicted trajectories of actors in an environment, where the predictions are differentiable with respect to parameters of a neural network of a motion predictor used to predict the trajectories. The analytical function may be used to determine candidate trajectories for the machine based on a predicted trajectory, to compute cost values for the candidate trajectories, and to select a reference trajectory from the candidate trajectories. For differentiability, a term of the analytical function may correspond to the predicted trajectory. A motion controller may use the reference trajectory to predict a control sequence for the machine using an analytical function trained to generate predictions that are differentiable with respect to at least one parameter of the analytical function used to compute the cost values.

公开(公告)号：US20250091605A1

公开(公告)日：2025-03-20

申请号：US18747265

申请日：2024-06-18

Applicant: NVIDIA Corporation

Inventor： Katie Z. Luo ,  Yue Wang ,  Xinshuo Weng ,  Marco Pavone ,  Yan Wang

IPC: B60W60/00 ,  G01C21/30

Abstract: In the context of autonomous driving, the recognition of lane topologies is required for the vehicle to make well-informed and prudent decisions such as lane changes, navigation through intricate intersections, and smooth merging. Current autonomous driving systems rely solely on sensor (e.g. camera) inputs to recognize lane topology. As a result, poor sensor data will have a direct negative impact on lane topology recognition. The present disclosure augments lane topology reasoning with a standard definition navigation map for use in autonomous driving applications.

公开(公告)号：US20240160913A1

公开(公告)日：2024-05-16

申请号：US18051114

申请日：2022-10-31

Applicant: NVIDIA Corporation

Inventor： Ryan Cosner ,  Yuxiao Chen ,  Karen Yan Ming Leung ,  Marco Pavone

IPC: G06N3/08

CPC classification number: G06N3/08

Abstract: In various examples, learning responsibility allocations for machine interactions is described herein. Systems and methods are disclosed that train a neural network(s) to generate outputs indicating estimated levels of responsibilities associated with interactions between vehicles or machines and other objects (e.g., other vehicles, machines, pedestrians, animals, etc.). In some examples, the neural network(s) is trained using real-world data, such as data representing scenes depicting actual interactions between vehicles and objects and/or parameters (e.g., velocities, positions, directions, etc.) associated with the interactions. Then, in practice, a vehicle (e.g., an autonomous vehicle, a semi-autonomous vehicle, etc.) may use the neural network(s) to generate an output indicating a proposed or estimated level of responsibility associated with an interaction between the vehicle and an object. The vehicle may then use the output to determine one or more controls for the vehicle to use when navigating.

公开(公告)号：US20240017745A1

公开(公告)日：2024-01-18

申请号：US17865344

申请日：2022-07-14

Applicant: NVIDIA Corporation

Inventor： Yulong Cao ,  Chaowei Xiao ,  Danfei Xu ,  Anima Anandkumar ,  Marco Pavone

IPC: B60W60/00 ,  B60W40/04 ,  B60W50/00

CPC classification number: B60W60/0027 ,  B60W40/04 ,  B60W50/0097 ,  B60W2554/4041 ,  B60W2554/4044

Abstract: Apparatuses, systems, and techniques to generate trajectory data for moving objects. In at least one embodiment, adversarial trajectories are generated to evaluate a trajectory prediction model and are based, at least in part, on a differentiable dynamic model.

公开(公告)号：US20240017743A1

公开(公告)日：2024-01-18

申请号：US18183582

申请日：2023-03-14

Applicant: NVIDIA Corporation

Inventor： Alec Farid ,  Sushant Veer ,  Boris Ivanovic ,  Karen Leung ,  Marco Pavone

IPC: B60W60/00 ,  B60W30/09

CPC classification number: B60W60/0016 ,  B60W60/00272 ,  B60W30/09

Abstract: In various examples, cost probability distributions corresponding to predicted locations of an object in an environment and potential locations for a machine in the environment and may be evaluated using corresponding observed costs corresponding to the machine and the object. The cost probability distributions may be evaluated based on comparing the observed costs to threshold values, which may be determined based on sampling a predicted cost function. A threshold value may be selected to provide false-positive rate and/or false-negative rate guarantees for anomaly detection. Control operations may be performed based on results of the evaluation of the cost probability distributions. For example, based on the results, a motion planner may reuse a planned trajectory for a future planning cycle (e.g., thereby avoiding re-planning computations) or generate and/or select a new planned trajectory (e.g., based at least on one or more anomalies being detected).

公开(公告)号：US20240273802A1

公开(公告)日：2024-08-15

申请号：US18422650

申请日：2024-01-25

Applicant: NVIDIA Corporation

Inventor： Yue Wang ,  Marco Pavone ,  Jiawei Yang

IPC: G06T15/00 ,  G06T15/06

CPC classification number: G06T15/005 ,  G06T15/06

Abstract: In various examples, frequency regularization and/or occlusion regularization techniques may be used to train Neural Radiance Fields (NeRF) to determine neural renderings based at least on sparse inputs in a way that reduces overfitting, underfitting, and/or occlusions. For example, while training a NeRF, a linearly increased frequency mask may be applied to regularize a visible frequency spectrum of training data based on training time steps. In examples, as training of the NeRF progresses, the visible frequency may be increased in a way that reduces the risk of overfitting and/or avoids underfitting. Additionally, the disclosed techniques may also include masking one or more density scores located within a threshold proximity of an origin of a ray to reduce floaters, walls, and other occlusions in the neural rendering output.