Abstract: Example embodiments relate to techniques for using vehicle image radar to estimate rain rate and other weather conditions. A computing device may receive radar data from a radar unit coupled to a vehicle. The radar data can represent the vehicle's environment. The computing device may use the radar data to determine a radar representation that indicates backscatter power and estimate, using a rain rate model, a rain rate for the environment based on the radar representation. The computing device may then control the vehicle based on the rain rate. In some examples, the computing device may provide the rain rate estimation and an indication of its current location to other vehicles to enable the vehicles to adjust routes based on the rain rate estimation.

Abstract: An example method and system for filtering point cloud data includes obtaining point cloud data from a LIDAR device. The point cloud data may include at least a first pulse-length range and a second pulse-length range. The first range may include one or more first-length pulses and the second range may include one or more second-length pulses. The method may further include filtering the point cloud data by determining respective magnitudes of each of the one or more first-length pulses and each of the one or more second-length pulses, comparing the magnitudes of the first-length pulses to a first threshold, comparing the magnitudes of the second-length pulses to a second threshold, and removing any pulses having a magnitude less than the respective thresholds. The method may further include determining, based on the filtered point cloud data, objects in an environment around the LIDAR.

Abstract: Example embodiments relate to lane adjustment techniques for slow lead agents. A vehicle computing system may use sensor data depicting the surrounding environment to detect when another vehicle is traveling in front of the vehicle at a speed that is less than a threshold minimum speed. If the other vehicle fails to increase speed above the minimum speed, the computing system may determine whether to change lanes to avoid the other vehicle based on speed data for other lanes. In some implementations, the computing system assigns penalties to lane segments surrounding the vehicle based on speed data for the different lane segments. For instance, the path finding system for the vehicle can use penalties and speed data to determine efficient routes that safely circumvent slow agents.

Abstract: Example systems and methods allow for reporting and sharing of information reports relating to driving conditions within a fleet of autonomous vehicles. One example method includes receiving information reports relating to driving conditions from a plurality of autonomous vehicles within a fleet of autonomous vehicles. The method may also include receiving sensor data from a plurality of autonomous vehicles within the fleet of autonomous vehicles. The method may further include validating some of the information reports based at least in part on the sensor data. The method may additionally include combining validated information reports into a driving information map. The method may also include periodically filtering the driving information map to remove outdated information reports. The method may further include providing portions of the driving information map to autonomous vehicles within the fleet of autonomous vehicles.

Abstract: Example embodiments relate to radar reflection filtering using a vehicle sensor system. A computing device may detect a first object in radar data from a radar unit coupled to a vehicle and, responsive to determining that information corresponding to the first object is unavailable from other vehicle sensors, use the radar data to determine a position and a velocity for the first object relative to the radar unit. The computing device may also detect a second object aligned with a vector extending between the radar unit and the first object. Based on a geometric relationship between the vehicle, the first object, and the second object, the computing device may determine that the first object is a self-reflection of the vehicle caused at least in part by the second object and control the vehicle based on this determination.

Abstract: The technology relates to controlling a vehicle based on a railroad light's activation status. In one example, one or more processors receive images of a railroad light. The one or more processors determine, based on the images of the railroad light, the illumination status of a pair of lights of the railroad light over a period of time as the vehicle approaches the railroad light. The one or more processors determine based on the illumination status of the pair of lights, a confidence level, wherein the confidence level indicates the likelihood the railroad light is active. The vehicle is controlled as it approaches the railroad light based on the confidence level.

Abstract: One example system comprises a light source configured to emit light. The system also comprises a waveguide configured to guide the emitted light from a first end of the waveguide toward a second end of the waveguide. The waveguide has an output surface between the first end and the second end. The system also comprises a plurality of mirrors including a first mirror and a second mirror. The first mirror reflects a first portion of the light toward the output surface. The second mirror reflects a second portion of the light toward the output surface. The first portion propagates out of the output surface toward a scene as a first transmitted light beam. The second portion propagates out of the output surface toward the scene as a second transmitted light beam.

Abstract: The present disclosure relates to systems and methods that utilize an internal optical path in lidar applications. An example method includes causing at least one light-emitter device to emit a plurality of light pulses toward a rotatable mirror. The rotatable mirror (i) reflects at least a first light pulse of the plurality of light pulses into an external environment; and (ii) reflects at least a second light pulse of the plurality of light pulses into an internal optical path. The method also includes receiving, by a photodetector, (i) a reflected light pulse comprising a reflection of the first light pulse caused by an object in the external environment; and (ii) the second light pulse received via the internal optical path. The internal optical path is defined at least in part by one or more internal reflectors that reflect the second light pulse toward the rotatable mirror such that the rotatable mirror reflects the second light pulse toward the photodetector.

Abstract: Aspects of the disclosure provide for automatically generating labels for sensor data. For instance first sensor data for a first vehicle is identified. The first sensor data is defined in both a global coordinate system and a local coordinate system for the first vehicle. A second vehicle is identified based on a second location of the second vehicle within a threshold distance of the first vehicle within the first timeframe. The second vehicle is associated with second sensor data that is further associated with a label identifying a location of an object, and the location of the object is defined in a local coordinate system of the second vehicle. A conversion from the local coordinate system of the second vehicle to the local coordinate system of the first vehicle may be determined and used to transfer the label from the second sensor data to the first sensor data.

Abstract: Example embodiments relate to self-reflection filtering techniques within radar data. A computing device may use radar data to determine a first radar representation that conveys information about surfaces in a vehicle's environment. The computing device may use a predefined model to generate a second radar representation that assigns predicted self-reflection values to respective locations of the environment based on the information about the surfaces conveyed by the first radar representation. The predefined model can enable a predefined self-reflection value to be assigned to a first location based on information about a surface positioned at a second location and a relationship between the first location and the second location. The computing device may then modify the first radar representation based on the predicted self-reflection values in the second radar representation and provide instructions to a control system of the vehicle based on modifying the first radar representation.

Abstract: An example device includes a mounting structure including a first material having a first coefficient of thermal expansion (CTE). The mounting structure includes a center portion and an outer portion. The device further includes a magnetic core for an electrical component that is coupled to the outer portion of the mounting structure. The magnetic core includes a second material having a second CTE. The magnetic core is split into a plurality of sections separated by spaces extending from the center portion to an outer edge of the outer portion. Each of the plurality of sections is separately coupled to the mounting structure, and each of the plurality of sections is connected to the electrical component.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, that obtain scene features in an environment that includes an autonomous vehicle, a first target agent, and a second target agent, and determines whether the first target agent is an emergency vehicle that is active at a current time point. In response to determining that the first target agent is an emergency vehicle that is active at the current time point, an input is generated from the scene features. The input can characterize the scene and indicate that the first target agent is an emergency vehicle that is active at the current time point. Also in response, the input can be processed using a machine learning model that is configured to generate a trajectory prediction output for the second target agent that characterizes predicted future behavior of the second target agent after the current time point.

Abstract: Disclosed herein are methods and apparatus for controlling autonomous vehicles utilizing maps that include visibility information. A map is stored at a computing device associated with a vehicle. The vehicle is configured to operate in an autonomous mode that supports a plurality of driving behaviors. The map includes information about a plurality of roads, a plurality of features, and visibility information for at least a first feature in the plurality of features. The computing device queries the map for visibility information for the first feature at a first position. The computing device, in response to querying the map, receives the visibility information for the first feature at the first position. The computing device selects a driving behavior for the vehicle based on the visibility information. The computing device controls the vehicle in accordance with the selected driving behavior.

Abstract: Example embodiments relate to techniques for enabling one or more systems of a vehicle (e.g., an autonomous vehicle) to request remote assistance to help the vehicle navigate in an environment. A computing device may be configured to receive a request for assistance from a vehicle. The request may include an image frame representative of a portion of an environment. The computing device may also be configured to initiate display of a graphical user interface to visually represent the image frame. Further, the computing device may determine a bounding region for the image frame. The bounding region may be associated with one or more objects in the image frame. Additionally, the computing device may be configured to receive, via the GUI, an input that includes an object identifier, and associate the object identifier with each of the one or more objects in the bounding region.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generated simulated sensor data. One of the methods includes obtaining a surfel map generated from sensor observations of a real-world environment and generating, for each surfel in the surfel map, a respective grid having a plurality of grid cells, wherein each grid has an orientation matching an orientation of a corresponding surfel, and wherein each grid cell within each grid is assigned a respective color value. For a simulated location within a simulated representation of the real-world environment, a textured surfel rendering is generated, including combining color information from grid cells visible from the simulated location within the simulated representation of the real-world environment.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for detecting regions of an environment. One of the methods includes receiving a representation of a scene in an environment; processing the representation using a center prediction neural network to generate: (i) features of the scene in the environment, and (ii) a respective center score corresponding to each of a plurality of locations in the environment; selecting, based on the respective center scores, one or more of the plurality of locations; and for each selected location: processing an input comprising the features of the scene in the environment and data specifying the selected location using a geometry prediction neural network to generate a geometry prediction that represents a geometry of the region that is centered at the selected location.

Abstract: An optical system may include a substrate and a plurality of silicon photomultipliers (SiPMs) monolithically integrated with the substrate. Each SiPM may include a plurality of single photon avalanche diodes (SPADs). The optical system also includes an aperture array having a plurality of apertures. The plurality of SiPMs and the aperture array are aligned so as to define a plurality of receiver channels. Each receiver channel includes a respective SiPM of the plurality of SiPMs optically coupled to a respective aperture of the plurality of apertures.

Abstract: Aspects and implementations of the present disclosure address shortcomings of the existing technology by enabling lidar-assisted segmentation and identification of particulate matter in autonomous vehicle (AV) applications, by: obtaining, by a sensing system of the AV, a plurality of return points, each return point having one or more velocity values and one or more coordinates of a reflecting region that reflects a signal emitted by the sensing system, identifying, in view of the one or more velocity values of each of a first set of the return points of the plurality of return points, that the first set of the return points is associated with a particulate matter in an environment of the AV, and causing a driving path of the AV to be determined in view of the particulate matter.

Abstract: A method includes obtaining multiple images captured by pixel sensors of an image sensor, analyzing, using neural network circuitry integrated in the image sensor, the multiple images for object detection, generating, for each of the multiple images using the neural network circuitry integrated in the image sensor, neural network output data related to results of the analysis of the multiple images for object detection, and transmitting, from the image sensor, the neural network output data for each of the multiple images and image data for a subset of the multiple images instead of image data of each of the multiple images.

Abstract: Example implementations are provided for an arrangement of co-aligned rotating sensors. One example device includes a light detection and ranging (LIDAR) transmitter that emits light pulses toward a scene according to a pointing direction of the device. The device also includes a LIDAR receiver that detects reflections of the emitted light pulses reflecting from the scene. The device also includes an image sensor that captures an image of the scene based on at least external light originating from one or more external light sources. The device also includes a platform that supports the LIDAR transmitter, the LIDAR receiver, and the image sensor in a particular relative arrangement. The device also includes an actuator that rotates the platform about an axis to adjust the pointing direction of the device.