A computer-implemented method for predicting properties of a plurality of objects in a vicinity of a vehicle includes multiple steps that can be carried out by computer hardware components. The method includes determining a grid map representation of road-users perception data, with the road-users perception data including tracked perception results and/or untracked sensor intermediate detections. The method also includes determining a grid map representation of static environment data based on data obtained from a perception system and/or a pre-determined map. The method further includes determining the properties of the plurality of objects based on the grid map representation of road-users perception data and the grid map representation of static environment data.

Provided are methods for semantic annotation of sensor data using unreliable map annotation inputs, which can include training a machine learning model to accept inputs including images representing sensor data for a geographic area and unreliable semantic annotations for the geographic area. The machine learning model can be trained against validated semantic annotations for the geographic area, such that subsequent to training, additional images representing sensor data and additional unreliable semantic annotations can be passed through the neural network to provide predicted semantic annotations for the additional images. Systems and computer program products are also provided.

Described herein are methods of generating learning data to facilitate de-biasing the labeled location of an object of interest within an image. Methods may include: receiving sensor data, where the sensor data is a first image; determining reference corner locations of an object in the first image using image processing; generating observed corner locations of the object in the first image from the determined reference corner locations; generating a bias transformation based, at least in part, on a difference between the reference corner locations and the observed corner locations of the object in the first image; receiving sensor data from another image sensor of a second image; receiving observed corner locations of an object in the second image from a user; and applying the bias transformation to the observed corner locations of the object in the second image to generate de-biased corners for the object in the second image.

Systems and methods for updating navigational maps based using at least one sensor are provided. In one aspect, a control system for an autonomous vehicle, includes a processor and a computer-readable memory configured to cause the processor to: receive output from at least one sensor located on the autonomous vehicle indicative of a driving environment of the autonomous vehicle, retrieve a navigational map used for driving the autonomous vehicle, and detect one or more inconsistencies between the output of the at least one sensor and the navigational map. The computer-readable memory is further configured to cause the processor to: in response to detecting the one or more inconsistencies, trigger mapping of the driving environment based on the output of the at least one sensor, update the navigational map based on the mapped driving environment, and drive the autonomous vehicle using the updated navigational map.

Certain aspects of the present disclosure provide techniques for performing piecewise pointwise convolution, comprising: performing a first piecewise pointwise convolution on a first subset of data received via a first branch input at a piecewise pointwise convolution layer of a convolutional neural network (CNN) model; performing a second piecewise pointwise convolution on a second subset of data received via a second branch input at the piecewise pointwise convolution layer; determining a piecewise pointwise convolution output by summing a result of the first piecewise pointwise convolution and a result of the second piecewise pointwise convolution; and providing the piecewise pointwise convolution output to a second layer of the CNN model.

A method of operation of a navigation system comprising: receiving multiple frames, including a first frame and a second frame, of images; detecting a traffic sign from the images between the first frame and the second frame based on a sign recognition model; extracting a first image from the first frame and a second image from the second frame; matching the traffic sign is the same in the first image and the second image based on a similarity model; generating a sign location of the traffic sign with an inertial measurement unit reading based on the first image and the second image; and generating a global coordinate for the sign location for displaying on a navigation map.

Aspects of the disclosure relate to determining perceptive range of a vehicle in real time. For instance, a static object defined in pre-stored map information may be identified. Sensor data generated by a sensor of the vehicle may be received. The sensor data may be processed to determine when the static object is first detected in an environment of the vehicle. A distance between the object and a location of the vehicle when the static object was first detected may be determined. This distance may correspond to a perceptive range of the vehicle with respect to the sensor. The vehicle may be controlled in an autonomous driving mode based on the distance.

Aspects of the disclosure provide for a method for identifying speed limit signs and controlling an autonomous vehicle in response to detected speed limit signs. The autonomous vehicle's computing devices identifies a speed limit sign in a vehicle's environment and a location and orientation corresponding to the speed limit sign. Then, the and orientation location of the speed limit sign is determined to not correspond to a pre-stored location and a pre-stored orientation of a speed limit sign that is pre-stored in map information. An effect zone of the speed limit sign is determined based on the location and orientation of the speed limit sign and characteristics of surrounding areas or other detected object before or after the speed limit sign. The autonomous vehicle's computing devices determines a response of the vehicle based on the determined effect zone, and controls the autonomous vehicle based on the determined response.

A system and method for determining a traffic sign violation are provided. The method may include obtaining a traffic rule corresponding to the traffic sign. The method may further include acquiring, by at least one camera, video data associated with a scene around a traffic sign. The video data may include a series of frames. The method may further include identifying the vehicle in the series of frames and determining whether the vehicle violates the traffic rule based on the series of frames. In response to the determination that the vehicle violates the traffic rule, the method may further include obtaining information of the vehicle and transmitting the information of the vehicle to a server.

Techniques for detecting attributes and/or gestures associated with pedestrians in an environment are described herein. The techniques may include receiving sensor data associated with a pedestrian in an environment of a vehicle and inputting the sensor data into a machine-learned model that is configured to determine a gesture and/or an attribute of the pedestrian. Based on the input data, an output may be received from the machine-learned model that indicates the gesture and/or the attribute of the pedestrian and the vehicle may be controlled based at least in part on the gesture and/or the attribute of the pedestrian. The techniques may also include training the machine-learned model to detect the attribute and/or the gesture of the pedestrian.