Hrl Laboratories, LLC

Gm Global Technolgoy Operations, Inc.

Rajan Bhattacharyya

Hyukseong Kwon

Amir M Rahimi

Michael D Howard

Michael J Daily

Suhas E Chelian

James Benvenuto

Amit Agarwal

Matthias Ziegler

Aruna Jammalamadaka

Matthew E Phillips

Kenji Yamada

Deepak Khosla

Kyungnam Kim

Praveen K Pilly

David J Huber

Soheil Kolouri

Vincent De Sapio

Dmitriy V Korchev

Youngkwan Cho

Aashish N Patel

Tiffany J Hwu

Kevin Lee

Alexie Pogue

Narayan Srinivasa

David W Payton

Kang-Yu Ni

Hankyu Moon

Ryan M Uhlenbrock

Rashmi N Sundareswara

Jaehoon Choe

Gavin D Holland

Samuel D Johnson

Chong Ding

Hyun (Tiffany) J Kim

Iman Mohammadrezazadeh

Michael J O'Brien

Christian Lebiere

Brian L Burns

Yuri Owechko

David L Allen

Qin Jiang

Jiejun Xu

Shinko Yuanhsien Cheng

Kevin R Martin

Terrell Nathan Mundhenk

Mike Howard

Giorgio A Ascoli

Rodolfo Valiente Romero

Jeffrey Lawrence Krichmar

Stephanie E Goldfarb

Scott Rad

Rachel Millin

Randall C O'Reilly

Jerry Vinokurov

Shanka R Rao

Srinivas Nedunuri

Penn Tasinga

Divya P Kulkarni

Joshua Lampkins

Meena Nagappan

Described is a system for generating and displaying decision recommendations. The system receives historical observations and courses of action corresponding to the historical observations. A conditional density between the historical observations and the courses of action is estimated, and the historical observations are clustered according to the conditional density to obtain observational clusters. Observational clusters are merged to determine a causal partition having corresponding causal features which map to various courses of action. A new observation is mapped to the causal partition and a ranking of courses of action for the new observation is computed. A visualization of causal features and a ranking of courses of actions corresponding to the new observation is provided to a display.

Decision recommendation via causal feature display

Described is a system for adapting to perception errors in object detection and recognition. The system receives, with a perception module, perception data from an environment proximate a mobile platform that reflects objects in the environment. Perception probes representing perception characteristics of object detections are generated from the perception data. Using the perception probes, spatial logic-based constraints and temporal logic-based constraints are generated. Spatial perception parameters are determined by solving an optimization problem using a set of the spatial logic-based constraints. Temporal perception parameters are determined by solving an optimization problem using a set of temporal logic-based constraints. The spatial perception parameters and the temporal perception parameters are combined to estimate a final perception parameter. The perception module is adjusted based on the final perception parameter.

Perception adaptation using probabilistic signal spatio-temporal logic system

A method for performing object detection during autonomous driving includes: performing 3D object detection in a 3D object detection segment; uploading an output of multiple sensors in communication with the 3D object detection segment to multiple point clouds; transferring point cloud data from the multiple point clouds to a Region Proposal Network (RPN); independently performing 2D object detection in a 2D object detector in parallel with the 3D object detection in the 3D object detection segment; and taking a given input image and simultaneously learning box coordinates and class label probabilities in a 2D object detection network operating to treat object detection as a regression problem.

3D object detection method using synergy of heterogeneous sensors for autonomous driving

Described is a system for action recognition error detection and correction using probabilistic signal temporal logic. The system is initiated by training an action recognition system to generate true positive (TP)/false positive (FP) axioms. Thereafter, the system ca be used to classify one or more actions in a video sequence as true action classifications by using the TP/FP axioms to remove false action classifications. With the remaining true classifications, a device can be controlled given the situation and relevant true classification.

System for action recognition error detection and correction using probabilistic signal temporal logic

A virtual lane estimation system includes a memory device, a sensor and a computer. The memory device is configured to store a road map that corresponds to a portion of a road ahead of a vehicle. The sensor is configured to observe a plurality of trajectories of a plurality of neighboring vehicles that traverse the portion of the road. The computer is configured to initialize a recursive self-organizing map as a plurality of points arranged as a two-dimensional grid aligned with the road map, train the points in the recursive self-organizing map in response to the trajectories, generate a directed graph that contains one or more virtual lanes through the road map in response to the points trained to the trajectories, and generate a control signal that controls navigation of the vehicle through the portion of the road in response to the virtual lanes in the directed graph.

Virtual lane estimation using a recursive self-organizing map

A method of using perception-inspired event generation for situation awareness for a vehicle, including receiving perception input data from a sensor of the vehicle and processing the perception input data to classify and generate parameters related to an external entity in a vicinity of the vehicle. The method includes generating a hierarchical event structure that classifies and prioritizes the perception input data by classifying the external entity into an attention zone and prioritizing the external entity within the attention zone according to a risk level value for the external entity. A higher risk level value indicates a higher priority within the attention zone. The method further includes developing a behavior plan for the vehicle based on the hierarchical event structure.

Methods for cognitive situation awareness using an attention-based event structure

Described is a system for evaluating and correcting perception errors in object detection and recognition. The system receives perception data from an environment proximate a mobile platform. Perception probes are generated from the perception data which describe perception characteristics of object detections in the perception data. For each perception probe, probabilistic distributions for true positive and false positive values are determined, resulting in true positive and false negative perception probes. Statistical characteristics of true positive perception probes and false positive perception probes are then determined. Based on the statistical characteristics, true positive perception probes are clustered. An axiom is generated to determine statistical constraints for perception validity for each perception probe cluster. The axiom is evaluated to classify the perception probes as valid or erroneous. Optimal perception parameters are generated by solving an optimization problem based on the axiom. The perception module is adjusted based on the optimal perception parameters.

Automated probabilistic axiom generation and incremental updates

A vehicle and a system and a method of operating the vehicle. The system includes a reasoning engine, an episodic memory, a resolver and a controller. The reasoning engine infers a plurality of possible scenarios based on a current state of an environment of the vehicle. The episodic memory determines a historical likelihood for each of the plurality of possible scenarios. The resolver selects a scenario from the plurality of possible scenarios using the historical likelihoods. The controller operates the vehicle based on the selected scenario.

System for extending functionality of hypotheses generated by symbolic/logic-based reasoning systems

Described is a system for analyzing time series data. A sequence of symbols is generated from a set of time series input data related to a moving vehicle using automatic segmentation. A grammar is extracted from the sequence of symbols, and the grammar is a subset of a probabilistic context-free grammar (PCFG). Using the grammar, time series input data can be analyzed, and a prediction of the vehicle's movement can be made. Vehicle operations for an autonomous vehicle are determined using the prediction.

Method for time series structure discovery

A method for updating a localization of an autonomous vehicle includes receiving perception input data from a sensor of the autonomous vehicle and receiving map data including road lane information in a vicinity of the autonomous vehicle. The method includes processing the perception input data to extract perceived road lane information including a perceived x position, a perceived y position, a perceived z position, a perceived lane type, a perceived lane color, and a perceived lane curvature and processing the map data to extract map road lane information including a map x position, a map y position, a map z position, a map lane type, a map lane color, and a map lane curvature. The method includes calculating a transformation matrix from the perceived road lane information and the map road lane information and updating the map data and a localization of the vehicle based on the transformation matrix.

Methods for multi-dimensional lane matching for autonomous vehicle localization

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.

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

Described is a system for detecting and correcting perception errors in a perception system. In operation, the system generates a list of detected objects from perception data of a scene, which allows for the generation of a list of background classes from backgrounds in the perception data associated with the list of detected objects. For each detected object in the list of detected objects, a closest background class is identified from the list of background classes. Vectors can then be used to determine a semantic feature, which is used to identify axioms. An optimal perception parameter is then generated, which is used to adjust perception parameters in the perception system to minimize perception errors.

Generation and usage of semantic features for detection and correction of perception errors

A control system of the autonomous vehicle may generate multiple possible behavior control movements based on the driving goal and the assessment of the vehicle environment. In doing so, the method and system selects one of the best behavior control, among the multiple possible movements, and the selection is based on the quantitative grading of its driving behavior.

Forward modeling for behavior control of autonomous vehicles

An ego vehicle includes decider modules and a grader module coupled to a resolver module. The decider modules generate trajectory decisions at a current time, generate a current two-dimensional slice of a flat space around the ego vehicle, generate future two-dimensional slices of the flat space by projecting the current two-dimensional slice of the flat space forward in time, and generate a three-dimensional state space by stacking the current two-dimensional slice and the future two-dimensional slices. The grader module generates rewards for the trajectory decisions based on a recent behavior of an ego vehicle. The resolver module selects a final trajectory decision for the ego vehicle from the trajectory decisions based on the three-dimensional state space and the rewards. The current two-dimensional slice includes a current ego vehicle location and current neighboring vehicle locations. The future two-dimensional slices include future ego vehicle locations and future neighboring vehicle locations.

Navigation trajectory using reinforcement learning for an ego vehicle in a navigation network

A method of path planning for a host vehicle includes: receiving host vehicle, environmental and obstacle information; calculating one or more projected host vehicle locations; computing a projected obstacle location for each obstacle; and determining a collision potential between each projected host vehicle location and each projected obstacle location. Until a maximum number of steps is reached, and while at least one projected host vehicle location has an associated collision potential below a collision threshold, the method further includes repeating the calculating, computing and determining steps.

System, method and controller for graph-based path planning for a host vehicle

Growing community of inventors

Sherman Oaks, CA, United States of America

Rajan Bhattacharyya