Abstract: A system may include a vehicle controller, comprising: a hardware platform, comprising a processor and a memory; and instructions encoded within the memory to: provide an autonomous vehicle control module with at least L4 vehicle control capability; select a semi-autonomous operating mode that uses a subset of the L4 vehicle control capability; and operate a vehicle in the semi-autonomous operating mode.

Abstract: Some autonomous vehicles rely on deep learning models to generate outputs that can be used to control the autonomous vehicle. Deep learning models can be trained to fit datasets collected in a particular environment. These deep learning models may not perform as well in a different environment, and new deep learning models may need to be created and trained. Training new models can be computationally expensive, and the amount of datasets collected in a new environment for training the new models may be limited. Various techniques involving sharing parameters (e.g., weights) between deep learning models can alleviate some of these challenges.

Abstract: Systems and methods for retrofitting a vehicle computing system dependent on a certain type of maps to operate with other type(s) of maps are provided. For example, a method performed by a vehicle may include receiving, from one or more sensors of the vehicle, first sensor data associated with a surrounding environment of the vehicle; receiving first map data of a first map type; and retrofitting a vehicle controller of the vehicle that is based on a second map type different from the first map type to operate on the first map data, where the retrofitting includes adapting the first map data using the first sensor data to generate second map data associated with the second map type; and determining, by the vehicle controller, an action for the vehicle based at least in part on the generated second map data.

Abstract: Systems and methods for retrofitting a light detection and ranging (LIDAR)-based vehicle computing system to operate with vision-based sensor data are provided. For example, a method implemented by a vehicle may include receiving, from one or more sensors of a first sensing modality at the vehicle, first sensor data associated with a surrounding environment of the vehicle; and retrofitting a vehicle controller of the vehicle that is based on a second sensing modality different from the first sensing modality to operate on the first sensor data, where the retrofitting includes generating second sensor data from the first sensor data based on the second sensing modality; and determining, by the vehicle controller, an action for the vehicle based at least in part on the generated second sensor data.

Abstract: An AV includes a 3D printer. The AV obtains a request to 3D print an object, e.g., based on identification of a feature associated with the AV or a user input. The AV can control the 3D printing process based on a motion of the AV that occurs during the 3D printing process. The AV may determine an influence of the motion on the 3D printing process and adjust the 3D printing process to compensate for the influence. Also, the AV may determine a motion during the 3D printing process to facilitate the 3D printing process, e.g., facilitate a movement of an item associated with the 3D printing process. The item may be at least a part of the 3D printer, a material from which the object is printed, or a portion of the object. A fleet management system can manage 3D printing services provided by a fleet of AVs.

Abstract: Systems and methods for identifying a precise stopping location using vehicle sensors and mechanical simulation. In particular, systems and methods are provided for using vehicle sensor information to understand vehicle surroundings, identifying an appropriate stopping position that either does not disrupt traffic (or minimally disrupts traffic), and identifying a path to the stopping position. Additionally, the suitability of the stopping position is determined by performing a simulation of the vehicle movement to the stopping position as well as a simulation of passenger behavior during vehicle exit. In some examples, the simulation is performed using a short-range relative map of the vehicle surroundings rather than using the driving map.

Abstract: An apparatus includes a processor and a machine-readable medium having program code executable by the processor to cause the apparatus to obtain a scan result based on a core sample, generate a set of contours of the scan result, and generate a set of object associations, wherein the set of object associations includes a first object association that is associated with a first contour from the set of contours and a second contour from the set of contours. The program code also generates a set of contour connections, wherein the set of contour connections includes a first contour connection that is based on the first object association. The program code also generates a reconstructed core having a three-dimensional space based on the set of contour connections, wherein the three-dimensional space includes a volume greater than the core sample.

Abstract: Users getting rides in AVs can engage in virtual environments through client devices (e.g., headsets, etc.) that project the virtual environments to the users. A user may perceive a virtual environment and engage in the virtual environment during a ride in an AV. The virtual environment includes a virtual scene generated based on the ride in the AV or an interaction of the user with another person in the real-world. The virtual scene includes a virtual representation of the user and other virtual objects. A virtual object may be generated based on an object, a behavior of the AV, or an activity of the user in the real-world. The user may perform various virtual activities in the virtual environment, such as communications with other users, digital interactive experiences, etc. A behavior of the AV (e.g., the AV's navigation) may be changed based on a virtual activity of the user.

Abstract: A model for controlling an AV can be trained with a privacy budget. A data set that includes sensor data collected by an AV from an environment around the AV is received. A privacy score of the data set is determined. The privacy score indicates a measurement of privacy information included in the data set. The privacy score may be compared with the privacy budget to determine whether the data set meets the privacy budget. After a determination that the data set does not meet the privacy budget, the data set is adjusted, e.g., by adjusting one or more objects captured by the sensor data. An example adjustment to an object may include removing or modifying a private feature of the object. The adjust data set has a privacy score that falls in the privacy budget and can be used as a privacy-protected data set to train the model.

Abstract: An operation of an AV in a scene is simulated. The AV includes sensors detecting the scene and generating sensor data. The sensor data can be input into a control model that outputs control signals, in accordance with which the AV operates. A learning cost, i.e., a cost of training or applying the control model, is determined. A performance of the AV during the simulation is evaluated. A simulation cost, i.e., a cost of running the simulation is determined. The control model can be optimized based on the learning cost and the performance of the AV. Settings of the sensors can be optimized based on the simulation cost and the performance of the AV. The optimization of the control model or the settings of the sensors can be done through reinforcement learning.

Abstract: A method for three dimensional visualization and manipulation of downhole data. A measured signal is received a downhole environment and a three dimensional virtualization of the measured signal is generated. A stereographic viewer displays the three dimensional virtualization of the measured signal. The three dimensional virtualization can be manipulated in response to an input from a user, thereby creating a manipulated three dimensional virtualization. The stereographic viewer can display the manipulated three dimensional virtualization.

Abstract: A disclosed downhole drilling system may include a drill bit electrically coupled to a pulse-generating circuit to generate electrical arcs between first and second electrodes during pulsed drilling operations, a sensor to record responses to electromagnetic or acoustic waves produced by the electrical arcs, and a sensor analysis system. The electrical arcs occur at different azimuthal locations between the electrodes. The sensor analysis system may obtain a plurality of measurements representing first responses recorded by the sensor during a pulsed drilling operation, generate a model of a source of the electrical arcs based on the measurements, obtain an additional measurement representing a second response recorded by the sensor during the operation, and determine a characteristic of a formation near the drill bit using an inversion based on the model and the additional measurement. The determined characteristic may be used to determine dip parameters or construct images of the formation.

Abstract: Disclosed herein are methods and system for formation fluid sampling. In at least some embodiments, the method includes pumping formation fluid from a public flow line of a downhole tool via a private flow line into a detachable sample chamber. The method also includes isolating the private flow line from the public flow line. The method also includes collecting measurements of the formation fluid in the private flow line. The method also includes associating the measurements with the detachable sample chamber.

Abstract: Machine learning model optimization systems and methods are disclosed. A system receives sensor data captured by one or more sensors of a vehicle during a first time period. The vehicle uses a first trained machine learning (ML) model for one or more decisions of a first decision type during the first time period. The system generates a second trained ML model at least in part by using the sensor data to train the second trained ML model. The system identifies an optimal trained ML model from a plurality of trained ML models. The plurality of trained ML models includes the first trained ML model and the second trained ML model. The system causes the vehicle to use the optimal trained ML model for one or more further decisions of the first decision type during a second time period after the first time period.

Abstract: Pipe]parameter determinations from electromagnetic logs can be improved, in accordance with various embodiments, by weighting signals with frequencies below a threshold associated with resolution degradation lower than signals with frequencies above the threshold. The threshold frequency may be determined based on a spatial resolution associated with the logging tool and a logging speed. Further embodiments are described.

Abstract: Aspects of the disclosed technology provide solutions for improving object detection and in particular, for improving object detection by an autonomous vehicle (AV) perceptions stack. In some aspects, a process of the disclosed technology can include steps for receiving first sensor data corresponding with an environment around the first vehicle, wherein the first sensor data represents one or more objects in the environment, receiving second sensor data corresponding with the environment around the first vehicle, and formulating a navigation route, by the first vehicle, based on the one or more occluded objects in the environment. Systems and machine-readable media are also provided.

Abstract: Disclosed are systems and techniques for managing an autonomous vehicle. In some aspects, an autonomous vehicle may obtain one or more radio frequency (RF) signals corresponding to a wireless device. In some cases, the autonomous vehicle may determine a location probability map associated with the wireless device based on the one or more RF signals. In some examples, the autonomous vehicle may adjust a behavior of the autonomous vehicle based on the location probability map associated with the wireless device.

Abstract: The subject disclosure relates to techniques for minimizing damage for collisions including an autonomous vehicle. A process of the disclosed technology can include predicting that a crash involving the autonomous vehicle is imminent, altering at least one operational parameter of the autonomous vehicle after predicting the crash is imminent, performing a first simulation on the autonomous vehicle, wherein the first simulation is a simulation of the autonomous vehicle taking a first action to minimize damage from the crash, and generating a first damage estimate for the first simulation.

Abstract: A method for three dimensional visualization and manipulation of downhole data. A measured signal is received a downhole environment and a three dimensional virtualization of the measured signal is generated. A stereographic viewer displays the three dimensional virtualization of the measured signal. The three dimensional virtualization can be manipulated in response to an input from a user, thereby creating a manipulated three dimensional virtualization. The stereographic viewer can display the manipulated three dimensional virtualization.

Abstract: The subject disclosure relates to techniques for enabling sharing of knowledge among a fleet of autonomous vehicles. A process of the disclosed technology can include generating an update for a continuous deep learning neural network on-board the autonomous vehicle based on driving scenarios encountered by the autonomous vehicle during its deployment and providing the update for the continuous deep learning neural network to additional vehicles in the fleet on autonomous vehicles, wherein the update for the continuous deep learning neural network is configured to be incorporated into a joint kernel for use by the additional vehicles in the fleet on autonomous vehicles.