Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for estimating a 3-D pose of an object of interest from image and point cloud data. In one aspect, a method includes obtaining an image of an environment; obtaining a point cloud of a three-dimensional region of the environment; generating a fused representation of the image and the point cloud; and processing the fused representation using a pose estimation neural network and in accordance with current values of a plurality of pose estimation network parameters to generate a pose estimation network output that specifies, for each of multiple keypoints, a respective estimated position in the three-dimensional region of the environment.

Abstract: The technology relates to localizing a vehicle. As one approach, a first LIDAR sensor scan data of an environment of the vehicle and localization data for the vehicle at a location where the first LIDAR sensor scan was captured are stored. Thereafter, the computing device is suspended and subsequently unsuspended. After the computing device is unsuspended, second LIDAR sensor scan data of the vehicle's environment is received. The first LIDAR sensor scan data is compared to the second LIDAR sensor scan data to determine whether the vehicle has moved. Based on the determination of whether the vehicle has moved from the location, the stored localization data is used to localize the vehicle. Other approaches are also described.

Abstract: The technology relates to detecting water or fouling on the exterior surface of a sensor cover, for instance due to precipitation or debris (fouling). Such objects on the sensor cover surface may degrade operation of the sensor, which can be problematic for vehicles operating in an autonomous driving mode. According to an aspect of the technology, a waveguide layer is provided on the sensor cover. A laser emits light at a selected wavelength along one side of the waveguide layer. The light waveform propagating along the waveguide layer is affected (distorted) by precipitation and/or fouling on the surface of this layer. A detector receives the distorted waveform. The system determines whether water and/or fouling is present based on the received waveform. This allows the system to determine whether to activate a cleaning module or to factor in the information when processing received sensor data.

Abstract: Example embodiments relate to a substrate integrated waveguide (SIW) with dual circular polarizations. An example SIW may include a dielectric substrate and a first metallic layer coupled to a top surface of the dielectric substrate with a through-hole extending through the dielectric substrate and the first metallic layer. The SIW also includes a dielectric layer coupled to a top surface of the first metallic layer. A second metallic layer is coupled to a top surface of the dielectric layer. The second metallic layer includes a non-conductive opening, a plurality of feeds with a first end in the non-conductive opening and a second end including a single-ended termination, and an impedance transformer. The SIW also includes a third metallic layer coupled to a bottom of the dielectric substrate, and a set of metallic via-holes proximate the non-conductive opening and coupling the second metallic layer to the third metallic layer.

Abstract: A light detection and ranging (LIDAR) system can emit light toward an environment and detect responsively reflected light to determine a distance to one or more points in the environment. The reflected light can be detected by a plurality of plurality of photodiodes that are reverse-biased using a high voltage. Signals from the plurality of reverse-biased photodiodes can be amplified by respective transistors and applied to an analog-to-digital converter (ADC). The signal from a particular photodiode can be applied to the ADC by biasing a respective transistor corresponding to the particular photodiode while not biasing transistors corresponding to other photodiodes. The gain of each photodiode/transistor pair can be controlled by adjusting the bias voltage applied to each photodiode using a digital-to-analog converter. The gain of each photodiode/transistor pair can be controlled based on the detected temperature of each photodiode.

Abstract: The technology includes a roof pod system for a vehicle configured to operate in one or more partly or fully autonomous self-driving modes. The roof pod system is arranged to sit above the roof of the vehicle, for instance at least 10-50 mm above the roof surface. The roof pod may be supported by a set of legs such as cross-rails. The roof pod system incorporates various sensors and related equipment to assist with self-driving operation. Some sensors may be arranged in the main housing of the roof pod, while others can be located in a dome-type structure extending above the main housing. A cabling harness assembly runs wiring and other links between the roof pod and the vehicle chassis.

Abstract: A control command is provided to a vehicle control module that identifies one or more vehicle actions to be performed by the vehicle control module to control a position of an autonomous vehicle (AV) with respect to an external environment. Whether one or more conditions pertaining to the position of the AV with respect to the external environment are satisfied is determined. Responsive to determining that the one or more conditions pertaining to the position of the AV with respect to the external environment are satisfied, a first instruction is provide to the vehicle control module that permits the vehicle control module to deviate from the one or more vehicle actions identified in the control command and to perform an energy saving action with respect to the AV.

Abstract: An apparatus includes an image sensor having a plurality of pixels that form regions of interest (ROIs), analog-to-digital converter (ADC) banks, and multiplexers. Each respective multiplexer is electrically connected to (i) a corresponding ADC bank and (ii) a corresponding subset of the ROIs. The apparatus also includes control circuitry configured to obtain a full-resolution image of an environment by electrically connecting, by way of the multiplexers, each respective ADC bank to the associated respective ROI. The control circuitry is also configured to select a particular ROI based on the full-resolution image and obtain a plurality of ROI images of the particular ROI by (i) electrically connecting, to the particular ROI, a first ADC bank associated with the particular ROI and a second ADC bank associated with another ROI and (ii) digitizing pixels of the particular ROI by way of parallel operation of the first and second ADC banks.

Abstract: Aspects of the disclosure relate to determining a next vehicle task for a vehicle of a fleet. Vehicle data from the vehicle, charger data about at least one charger, and demand data may be received and used to determine the next vehicle task. The vehicle may be directed to the next vehicle task. Determining a next vehicle task may further be based on predictions made using the vehicle data, the charger data, and the demand data. Heuristics may also be used in determining a next vehicle task.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for using a surfel map to generate long range localization. One of the methods includes obtaining, for a particular location of a vehicle having a camera and a detection sensor, surfel data including a plurality of surfels. Each surfel in the surfel data has a respective location and corresponds to a different respective detected surface in an environment. Image data captured by the camera is obtained. It is determined that a region of interest for detecting objects for a vehicle planning process is outside a detectable region for the detection sensor. In response, it is determined that the image data for the region of interest matches surfel color data for the surfels corresponding to the region of interest. In response, the vehicle planning process is performed with the region of interest designated as having no unexpected objects.

Abstract: One example method involves generating a calibration control signal that causes an actuator to rotate a first platform at least one complete rotation about an axis. The method also involves receiving encoder output signals. The encoder output signals are indicative of angular positions of the first platform about the axis. The method also involves receiving sensor output signals from an orientation sensor mounted on the first platform. The sensor output signals are indicative of a rate of change to an orientation of the orientation sensor. The method also involves determining calibration data based on given sensor output signals received from the orientation sensor during the at least one complete rotation. The calibration data is for mapping the encoder output signals to calibrated measurements of the angular positions of the first platform about the axis.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for detecting foreign objects using a surfel map. One of the methods includes receiving sensor data representing reflection characteristics of electromagnetic waves in an operating environment of an autonomous vehicle. A surfel map representing a portion of the operating environment of the autonomous vehicle is used to determine that the reflection data includes one or more mismatched reflection characteristics for a particular surfel in the surfel map. In response, a foreign object is detected at a location corresponding to the particular surfel having the one or more mismatched reflection characteristics.

Abstract: Systems and methods for extrinsic calibration of vehicle-mounted sensors are provided. One example method involves obtaining first sensor data collected by a first sensor and a second sensor while a vehicle is aligned in a first yaw direction. The method also involves obtaining second sensor data collected by the first sensor and the second sensor while the vehicle is aligned in a second yaw direction. The method also involves determining, based on the first sensor data and the second sensor data, (i) first pitch and roll misalignments of the first sensor relative to the vehicle and (ii) second pitch and roll misalignments of the second sensor relative to the first sensor. The method also involves determining third pitch and roll misalignments of the second sensor relative to the vehicle based on (i) the first pitch and roll misalignments and (ii) the second pitch and roll misalignments.

Abstract: The technology relates to determining a source of a horn honk or other noise in an environment around one or more self-driving vehicles. Aspects of the technology leverage real-time information from a group of self-driving vehicles regarding received acoustical information. The location and pose of each self-driving vehicle in the group, along with the precise arrangement of acoustical sensors on each vehicle, can be used to triangulate or otherwise identify the actual location in the environment for the origin of the horn honk or other sound. Other sensor information, map data, and additional data can be used narrow down or refine the location of a likely noise source. Once the location and source of the noise is known, each self-driving vehicle can use that information to modify current driving operations and/or use it as part of a reinforcement learning approach for future driving situations.

Abstract: Example embodiments relate to selective deactivation of light emitters for interference mitigation in light detection and ranging (lidar) devices. An example method includes deactivating one or more light emitters within a lidar device during a firing cycle. The method also includes identifying whether interference is influencing measurements made by the lidar device. Identifying whether interference is influencing measurements made by the lidar device includes determining, for each light detector of the lidar device that is associated with the one or more light emitters deactivated during the firing cycle, whether a light signal was detected during the firing cycle.

Abstract: Example embodiments relate to methods of increasing a temperature of a computer module to start the computer at environmental temperatures below a threshold temperature. An example embodiment includes receiving, at one or more computing components thermally coupled to a main computer via a liquid-cooled plate, a set of program instructions. The method can also include running the set of program instructions on at least one computing component. Running the set of program instructions on the computing component can generate heat that flows to the main computer via the liquid-cooled plate. The method can additionally include detecting, from at least one thermal sensor coupled to the liquid-cooled plate, a temperature reading indicative of a temperature of the main computer. The method can further include determining that the temperature reading has reached a predetermined temperature threshold and based on the temperature reading reaching the predetermined temperature threshold, powering on the main computer.