Abstract: In one example implementation according to aspects of the present disclosure, a computer-implemented method includes capturing a plurality of images at a camera associated with a vehicle and storing image data associated with the plurality of images to a memory. The method further includes dispatching vehicle perception tasks to a plurality of processing elements of an accelerator in communication with the memory. The method further includes performing, by at least one of the plurality of processing elements, the vehicle perception tasks for the vehicle perception using a neural network, wherein performing the vehicle perception tasks comprises performing an activation bypass for values below a first threshold, and performing weight pruning of synapses and neurons of the neural network based at least in part on a second threshold. The method further includes controlling the vehicle based at least in part on a result of performing the vehicle perception tasks.

Abstract: Examples of techniques for using fixed-point quantization in deep neural networks are disclosed. In one example implementation according to aspects of the present disclosure, a computer-implemented method includes capturing a plurality of images at a camera associated with a vehicle and storing image data associated with the plurality of images to a memory. The method further includes dispatching vehicle perception tasks to a plurality of processing elements of an accelerator in communication with the memory. The method further includes performing, by at least one of the plurality of processing elements, the vehicle perception tasks for the vehicle perception using a neural network, wherein performing the vehicle perception tasks comprises quantizing a fixed-point value based on an activation input and a synapse weight. The method further includes controlling the vehicle based at least in part on a result of performing the vehicle perception tasks.

Abstract: A method of on-line diagnostic and prognostic assessment of an autonomous vehicle perception system includes detecting, via a sensor, a physical parameter of an object external to the vehicle. The method also includes communicating data representing the physical parameter via the sensor to an electronic controller. The method additionally includes comparing the data from the sensor to data representing the physical parameter generated by a geo-source model. The method also includes comparing results generated by a perception software during analysis of the data from the sensor to labels representing the physical parameter from the geo-source model. Furthermore, the method includes generating a prognostic assessment of a ground truth for the physical parameter of the object using the comparisons of the sensor data to the geo-source model data and of the software results to the geo-source model labels. A system for on-line assessment of the vehicle perception system is also disclosed.

Abstract: A system and method to perform detection based on sensor fusion includes obtaining data from two or more sensors of different types. The method also includes extracting features from the data from the two or more sensors and processing the features to obtain a vector associated with each of the two or more sensors. The method further includes concatenating the two or more vectors obtained from the two or more sensors to obtain a fused vector, and performing the detection based on the fused vector.

Abstract: Described herein are systems, methods, and computer-readable media for generating and training a high precision low bit convolutional neural network (CNN). A filter of each convolutional layer of the CNN is approximated using one or more binary filters and a real-valued activation function is approximated using a linear combination of binary activations. More specifically, a non-1×1 filter (e.g., a k×k filter, where k>1) is approximated using a scaled binary filter and a 1×1 filter is approximated using a linear combination of binary filters. Thus, a different strategy is employed for approximating different weights (e.g., 1×1 filter vs. a non-1×1 filter). In this manner, convolutions performed in convolutional layer(s) of the high precision low bit CNN become binary convolutions that yield a lower computational cost while still maintaining a high performance (e.g., a high accuracy).

Abstract: An adaptive parallel imaging processing system in a vehicle is provided. The system may include, but is not limited to, a plurality of processors and a resource management system including, but not limited to, an execution monitor, the execution monitor configured to calculate an average utilization of each of the plurality of processors over a moving window, and a service scheduler controlling a request queue for each of the plurality of processors, the service scheduler scheduling image processing tasks in the respective request queue for the each of the plurality of processors based upon the average utilization of each of the plurality of processors, the capabilities of each of the plurality of processors, and a priority associated with each image processing task, wherein an autonomous vehicle control system is configured to generate the instructions to control the at least one vehicle system based upon the processed image processing tasks.

Abstract: A method and apparatus for updating an application are disclosed. According to the method, an update tag corresponding to a target application is pushed to a terminal when determining that the target application is required to be updated, the target application being an application downloaded and installed in the background of the terminal. An acquisition request sent by the terminal based on the update tag is received, the acquisition request being used for requesting to acquire target update content corresponding to the target application. The target update content is pushed to the terminal based on the update tag, such that the terminal opens the target application of a latest version by loading the target update content.

Abstract: In various embodiments, methods, systems, and vehicles are provided for calibrating vehicle cameras. In certain embodiments, a vehicle includes a camera, a memory, and a processor. The camera is disposed onboard the vehicle, and is configured to generate a camera image in which an object is detected onboard the vehicle. The memory is configured to store map data relating to the detected object. The processor is disposed onboard the vehicle, and is configured to perform an initial projection using the map data and initial values of calibration parameters for the camera; and update values of the calibration parameters using a comparison of the initial projection with the camera image.

Abstract: One general aspect includes a memory configured to include a program and a processor configured to execute the program, where the program enables the processor to: after a first vehicle ingress/egress event, activate a sensor to capture a reference image of a portion of a vehicle body; after a second vehicle ingress/egress event, activate the sensor to capture a test image of the portion of the vehicle body; determine whether the test image includes one or more objects not found in the reference image; and release the door from the body or produce a notification or prevent vehicle movement or some combination thereof, based on the determination of whether the test image includes one or more objects not found in the reference image.

Abstract: Systems and methods are provided for controlling an autonomous vehicle (AV). A scene understanding module of a high-level controller selects a particular combination of sensorimotor primitive modules to be enabled and executed for a particular driving scenario from a plurality of sensorimotor primitive modules. Each one of the particular combination of the sensorimotor primitive modules addresses a sub-task in a sequence of sub-tasks that address a particular driving scenario. A primitive processor module executes the particular combination of the sensorimotor primitive modules such that each generates a vehicle trajectory and speed profile. An arbitration module selects one of the vehicle trajectory and speed profiles having the highest priority ranking for execution, and a vehicle control module processes the selected one of vehicle trajectory and speed profiles to generate control signals used to execute one or more control actions to automatically control the AV.

Abstract: Systems and methods are provided for controlling an autonomous vehicle (AV). A map generator module processes sensor data to generate a world representation of a particular driving scenario (PDS). A scene understanding module (SUM) processes navigation route data, position information and a feature map to define an autonomous driving task (ADT), and decomposes the ADT into a sequence of sub-tasks. The SUM selects a particular combination of sensorimotor primitive modules (SPMs) to be enabled and executed for the PDS. Each one of the SPMs addresses a sub-task in the sequence. A primitive processor module executes the particular combination of the SPMs such that each generates a vehicle trajectory and speed (VTS) profile.

Abstract: Systems and methods are provided for controlling an autonomous vehicle (AV). A feature map generator module generates a feature map (FM). Based on the FM, a perception map generator module generates a perception map (PM). A scene understanding module selects from a plurality of sensorimotor primitive modules (SPMs), based on the FM, a particular combination of SPMs to be enabled and executed for the particular driving scenario (PDS). Each SPM maps information from either the FM or the PM to a vehicle trajectory and speed profile (VTSP) for automatically controlling the AV to cause the AV to perform a specific driving maneuver. Each one of the particular combination of the SPMs addresses a sub-task in a sequence of sub-tasks that address the PDS. Each of the particular combination of the SPMs are retrieved from memory and executed to generate a corresponding VTSP.

Abstract: A system and method for determining an object location by using map information is disclosed. The method includes receiving, by a controller, an image data of a scene. An object is located within the scene. The method also includes determining, by the controller, a location of the object based on the image data of the scene. The method can also include receiving, by the controller, a map information. The map information includes at least one path information. The method also includes determining an association between the location of the object based on the image data and the map information of the scene. The method also includes determining a 3-dimensional location of the object based on the determined association.

Abstract: A method and apparatus that detect wheel misalignment are provided. The method includes predicting a self-aligning torque parameter based on a regression model determined from a dataset including one or more from among a steering wheel angle parameter, a speed parameter, a torsion bar torque parameter, a lateral acceleration parameter, and a power steering torque parameter, comparing a measured self-aligning torque parameter and the predicted self-aligning torque parameter, and outputting a wheel alignment condition indicating whether the wheel alignment is proper if the self-aligning torque parameter and the predicted self-aligning torque parameter are within a predetermined value based on the comparing.

Abstract: An apparatus and method that detect a wheel alignment condition are provided. The method includes receiving a dataset comprising one or more from among a steering wheel angle parameter, a speed parameter, a lateral acceleration parameter, a self-aligning torque parameter and a power steering torque parameter, normalizing the received dataset, analyzing the normalized dataset according to a model for determining a wheel alignment condition, and outputting a value indicating whether the wheel alignment is within a predetermined value based on the model.

Abstract: A method of controlling a vehicle includes determining a current operating situation of the vehicle, and identifying a subset of a plurality of sensors of the vehicle needed to provide data to enable a vehicle control function for the current operating situation of the vehicle. A remainder of the plurality of sensors is disengaged to reduce electric energy usage by the vehicle while the vehicle is operating in the current operating situation of the vehicle. A sampling rate for the selected subset of sensors may be reduced to further reduce energy usage of the vehicle. Additionally, an energy reduction processing strategy may be implemented to reduce a processor frequency or a voltage of a computing device used to provide the vehicle control function to further reduce energy usage of the vehicle.

Abstract: Methods and apparatus are provided for controlling an autonomous vehicle. A sensor fusion system with a sensor system for providing environment condition information and a convolutional neural network (CNN) is provided. The CNN includes a receiving interface configured to receive the environment condition information from the sensor system, a common convolutional layer configured to extract traffic information from the received environment condition information, and a plurality of fully connected layers configured to detect objects belonging to different object classes based on the extracted traffic information, wherein the object classes include at least one of a road feature class, a static object class, and a dynamic object class.

Abstract: A system and method to perform detection based on sensor fusion includes obtaining data from two or more sensors of different types. The method also includes extracting features from the data from the two or more sensors and processing the features to obtain a vector associated with each of the two or more sensors. The method further includes concatenating the two or more vectors obtained from the two or more sensors to obtain a fused vector, and performing the detection based on the fused vector.

Abstract: A system and method to perform inter-sensor learning obtains a detection of a target based on a first sensor. The method also includes determining whether a second sensor with an overlapping detection range with the first sensor also detects the target, and performing learning to update a detection algorithm used with the second sensor based on the second sensor failing to detect the target.

Abstract: A method of identifying a condition of a road surface includes capturing at least a first image of the road surface with a first camera, and a second image of the road surface with a second camera. The first image and the second image are tiled together to form a combined tile image. A feature vector is extracted from the combined tile image using a convolutional neural network, and a condition of the road surface is determined from the feature vector using a classifier.