Abstract: The technology relates to fiducials for calibrating or validating a vehicle's sensors. One or more computing devices receive a plurality of data sets captured by the sensors from the plurality of fiducials. Each of the data sets corresponds to a respective sensor of the plurality of sensors. The one or more computing devices determine whether data in each of the plurality of data sets contains expected data representative of a respective fiducial of the plurality of fiducials. For each of the respective sensors, the one or more computing devices output a calibration status or a validation status associated with each data set that contains the expected data representative of the respective fiducial.

Abstract: Aspects of the technology relate to exception handling for a vehicle. For instance, a current trajectory for the vehicle and sensor data corresponding to one or more objects may be received. Based on the received sensor data, projected trajectories of the one or more objects may be determined. Potential collisions with the one or more objects may be determined based on the projected trajectories and the current trajectory. One of the potential collisions that is earliest in time may be identified. Based on the one of the potential collisions, a safety-time-horizon (STH) may be identified. When a runtime exception occurs, before performing a precautionary maneuver to avoid a collision, waiting no longer than the STH for the runtime exception to resolve.

Abstract: In another example, a method is provided that includes transmitting a first radar signal by a first channel of a plurality of channels of a radar unit. The method also includes receiving first radar reflections of the first radar signal by at least one reception antenna of the radar unit. Additionally, the method includes transmitting a second radar signal by the first channel. The method further includes receiving second radar reflections of the second radar signal by at least one reception antenna. Yet further, the method includes processing the received radar reflections from the first and second radar reflections to determine a platform movement parameter. Additionally, a virtual array may be constructed and radar signals received at common location of the virtual array may be used to determine a platform movement parameter. Moreover, the method includes operating the radar unit based on an offset determined from the platform movement parameter.

Abstract: Methods, systems, and apparatus for a stray-light testing apparatus. In one aspect, the apparatus includes an optical assembly including a spatially extended light source and one or more optical elements arranged to direct light from the spatially extended light source along an optical path, a moveable frame supporting the optical assembly including one or more adjustable alignment features for guiding positioning of the stray-light testing apparatus relative to an onboard camera on a vehicle, and a shrouding mechanism attached to the frame and positioned on the frame such that, when the stray-light testing apparatus is aligned relative to the onboard camera on the vehicle and the optical path of the optical assembly is within the field of view of the onboard camera, ambient light exposure for the onboard camera is below a threshold.

Abstract: Aspects of the disclosure relate to determining whether a feature of map information. For example, data identifying an object detected in a vehicle's environment and including location coordinates is received. This information is used to identify a corresponding feature from pre-stored map information based on a map location of the corresponding feature. The corresponding feature is defined as a curve and associated with a tag identifying a type of the corresponding object. A tolerance constraint is identified based on the tag. The curve is divided into two or more line segments. Each line segment has a first position. The first position of a line segment is changed in order to determine a second position based on the location coordinates and the tolerance constraint. A value is determined based on a comparison of the first position to the second position. This value indicates a likelihood that the corresponding feature has changed.

Abstract: Methods and systems are disclosed for determining sensor degradation by actively controlling an autonomous vehicle. Determining sensor degradation may include obtaining sensor readings from a sensor of an autonomous vehicle, and determining baseline state information from the obtained sensor readings. A movement characteristic of the autonomous vehicle, such as speed or position, may then be changed. The sensor may then obtain additional sensor readings, and second state information may be determined from these additional sensor readings. Expected state information may be determined from the baseline state information and the change in the movement characteristic of the autonomous vehicle. A comparison of the expected state information and the second state information may then be performed. Based on this comparison, a determination may be made as to whether the sensor has degraded.

Abstract: Machine-learning models are described detecting the signaling state of a traffic signaling unit. A system can obtain an image of the traffic signaling unit, and select a model of the traffic signaling unit that identifies a position of each traffic lighting element on the unit. First and second neural network inputs are processed with a neural network to generate an estimated signaling state of the traffic signaling unit. The first neural network input can represent the image of the traffic signaling unit, and the second neural network input can represent the model of the traffic signaling unit. Using the estimated signaling state of the traffic signaling unit, the system can inform a driving decision of a vehicle.

Abstract: The technology employs a holistic approach to passenger pickups and other wayfinding situations. This includes identifying where passengers are relative to the vehicle and/or the pickup location. Information synthesis from different sensors, agent behavior prediction models, and real-time situational awareness are employed to identify the likelihood that the passenger to be picked up is at a given location at a particular point in time, with sufficient confidence. The system can provide adaptive navigation by helping passengers understand their distance and direction to the vehicle, for instance using various cues via an app on the person's device. Rider support tools may be provided, which enable a remote agent to interact with a customer via that person's device, such as using the camera on the device to provide wayfinding support to enable the person to find their vehicle. Ride support may also use sensor information from the vehicle when providing wayfinding support.

Abstract: An article of manufacture includes a support structure including a cladding material and defining therein a plurality of substantially parallel cores. The article also includes a plurality of conically-shaped spikes protruding from a first side of the support structure. Each respective conically-shaped spike of the plurality of conically-shaped spikes includes a core material (i) extending through a corresponding core of the plurality of substantially parallel cores and (ii) comprising an axial protrusion that protrudes axially from the cladding material at the first side of the support structure. The axial protrusion of the core material is tapered to form the respective conically-shaped spike. The article also includes a refractory metal layer coating at least a portion of each respective conically-shaped spike and one or more electrodes connected to the refractory metal layer and configured to apply a voltage to the refractory metal layer.

Abstract: Aspects of the disclosure provide for localizing a vehicle. In one instance, a weather condition in which the vehicle is currently driving may be identified. A plurality of sensor inputs including intensity information, elevation information, and radar sensor information may be received. For each of the plurality of sensor inputs, an alignment score is determined by comparing the intensity information, elevation information, and radar sensor information to a corresponding pre-stored image for each of the intensity information, the elevation information, and the radar sensor information. A set of weights for the plurality of sensor inputs may be determined based on the identified weather condition. The alignment scores may then be combined using the set of weights in order to localize the vehicle.

Abstract: A camera includes a camera focus adjustment device, a lens, and an image sensor coupled to the camera focus adjustment device. The camera focus adjustment device includes a flexure structure. The flexure structure includes an outer framework of structural members continuously interconnected by flexure notch hinges. The flexure structure also includes two inner structural members oriented in parallel and extending from the outer framework of structural members. A gap is between the two inner structural members. The camera focus adjustment device also includes a piezoelectric material within the gap and a pair of wedges within the gap. The pair of wedges is affixed to the piezoelectric material and to one inner structural member of the two inner structural members. Based on temperature-based piezoelectric activity associated with the piezoelectric material, the camera focus adjustment device is operable to move the image sensor relative to the lens.

Abstract: Example embodiments described herein involve a system for testing a light-emitting module. The light-emitting module may include a mounting platform configured to hold a light-emitting module for a camera. The mounting platform may also be configured to rotate. The system may further include a housing holding a plurality of photodiodes arranged in an array over at least a 90 degree arc of a hemisphere. The system may also include a controller configured to control the photodiodes and the rotation of the mounting platform.

Abstract: The present disclosure is directed to an autonomous vehicle having a vehicle control system. The vehicle control system includes an image processing system. The image processing system receives an image that includes a plurality of image portions. The image processing system also calculates a score for each image portion. The score indicates a level of confidence that a given image portion represents an illuminated component of a traffic light. The image processing system further identifies one or more candidate portions from among the plurality of image portions. Additionally, the image processing system determines that a particular candidate portion represents an illuminated component of a traffic light using a classifier. Further, the image processing system provides instructions to control the autonomous vehicle based on the particular candidate portion representing an illuminated component of a traffic light.

Abstract: This technology relates to a system for preventing particle buildup on a sensor housing. The system may include a sensor housing including a first surface, a motor, and a spoiler edge. The motor may be configured to rotate the sensor housing around an axis. The spoiler edge may be positioned adjacent to the first surface and extended away from the first surface perpendicular to the axis of rotation of the sensor housing.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for associating a new measurement of an object surrounding a vehicle with a maintained track. One of the methods includes receiving an object track for a particular object, receiving a new measurement characterizing a new object at a new time step, and determining whether the new object is the same as the particular object, comprising: generating a representation of the new object at the new and preceding time steps; generating a representation of the particular object at the new and preceding time steps; processing a first network input comprising the representations using a first neural network to generate an embedding of the first network input; and processing the embedding of the first network input using a second neural network to generate a predicted likelihood that the new object and the particular object are the same.

Abstract: Aspects of the disclosure provide for advanced trip planning for an autonomous vehicle service. For instance, an example method may include determining a potential pickup location for a user, determining a set of potential destination locations for a user, and determining a set of potential trips. For each potential trip a vehicle of a fleet of autonomous vehicles of the service may be assigned and trip information, including an estimated time of arrival for the assigned vehicle of the potential trip to reach the destination location of the potential trip, may be determined. The trip information for each potential trip may be provided for display to the user. Thereafter, confirmation information identifying one of the set of potential trips may be received, and the assigned vehicle for one first of the set of potential trips may be dispatched to pick up the user.

Abstract: Aspects of the disclosure provide for evaluation of a planned trajectory for an autonomous vehicle. For instance, for each of a plurality of objects, a predicted trajectory may be received. The planned trajectory may identify locations and times that the vehicle will be at those locations. For each of the plurality of objects, a grid including a plurality of cells may be generated. Occupancy of each grid for each of the plurality of objects may be determined based on the predicted trajectories. A cell of each grid which will be occupied by the vehicle at a location and time of the planned trajectory may be identified. The planned trajectory may be evaluated based on whether any identified cell is occupied by any of the plurality of objects at the time.