Techniques described herein are directed to comparing, using a machine-trained model, neural network activations associated with data representing a simulated environment and activations associated with data representing real environment to determine whether the simulated environment is causes similar responses by the neural network, e.g., a detector. If the simulated environment and the real environment do not activate the same way (e.g., the variation between neural network activations of real and simulated data meets or exceeds a threshold), techniques described herein are directed to modifying parameters of the simulated environment to generate a modified simulated environment that more closely resembles the real environment.

Techniques for estimating a height range of an object in an environment are discussed herein. For example, a sensor, such as a lidar sensor, can capture three-dimensional data of an environment. The sensor data can be associated with a two-dimensional representation. A ground surface can be removed from the sensor data, and clustering techniques can be used to cluster remaining sensor data provided in a two-dimensional representation to determine object(s) represented therein. A height of a sensor object can be represented as a first height based on an extent of the sensor data associated with the object and can be represented as a second height based on beam spreading aspects of the sensor data and/or sensor data associated with additional objects. Thus, a minimum and/or maximum height of an object can be determined in a robust manner. Such height ranges can be used to control an autonomous vehicle.

A method for identifying a road condition of a road. A piece of road condition information representing the road condition is determined using a noise level detected by at least one ultrasonic sensor of a vehicle and a bottom echo detected from a road surface in the area of the vehicle.

Methods, systems, and products for resolving level ambiguity for radar systems of autonomous vehicles may include detecting a plurality of objects with a radar system. Each first detected object may be associated with an existing tracked object based on a first position thereof. First tracked object data based on a first height determined for each first detected object may be stored. The first height may be based on the position of the detected object, the existing tracked object, and a tile map. Second tracked object data based on a second height determined for each second detected object not associated with the existing tracked object(s) may be stored. The second height may be based on a position of each second detected object, a vector map, and the tile map. A command to cause the autonomous vehicle to perform at least one autonomous driving operation may be issued.

A system and method for detection of an aerial drone in an environment includes a baseline of geo-mapped sensor data in a temporal and location indexed database formed by (i) using at least one sensor to receive signals from the environment and converting into digital signals for further processing; (ii) deriving time delays, object signatures, Doppler shifts, reflectivity, and/or optical characteristics from the received signals; (iii) geo-mapping the environment using GNSS and the sensor data; and (iv) logging sensor data over a time interval, for example 24 hours to 7 days. Live sensor data can be then be monitored and signature data can be derived by computing at least one parameter such as direction and signal strength. The live data is continuously or periodically compared to the baseline data to identify a variance, if any, which may be indicative of a detection event.

In some implementations, a method of verifying operation of a sensor is provided. The method includes causing a sensor to obtain sensor data at a first time, wherein the sensor obtains the sensor data by emitting waves towards a detector. The method also includes determining that the detector has detected the waves at a second time. The method further includes receiving the sensor data from the sensor at a third time. The method further includes verifying operation of the sensor based on at least one of the first time, the second time, or the third time.

An example system may include a processor and a non-transitory machine-readable storage medium storing instructions executable by the processor to generate, from data collected by a sensor, a model of an area within which a mmWave sensor is to be N utilized for presence detection; shape, based on the model, a detection field of view of the mmWave sensor to be contained within the N area; and perform the presence detection within the area utilizing the shaped detection field of view of the mmWave sensor.

An autonomous vehicle (AV) includes a radar sensor system and a computing system that computes velocities of an object in a driving environment of the AV based upon radar data that is representative of radar returns received by the radar sensor system. The AV can be configured to compute a first velocity of the object based upon first radar data that is representative of the radar return from a first time to a second time. The AV can further be configured to compute a second velocity of the object based upon second radar data that includes at least a portion of the first radar data and further includes additional radar data representative of a radar return received subsequent to the second time. The AV can further be configured to control one of a propulsion system, a steering system, or a braking system to effectuate motion of the AV based upon the computed velocities.

Certain aspects of the present disclosure provide techniques for sensor calibration. First sensor data is received from a first sensor and second sensor data is received from a second sensor, where the first sensor data and the second sensor data each indicate detected objects in a space. The first sensor data is transformed using a first transformation profile to convert the first sensor data to a coordinate frame of the second sensor data. The first transformation profile is refined based on a difference between the transformed first sensor data and the second sensor data.

Commercial personal mobile vehicles (PMVs) managed by a fleet management system are sometimes equipped with a radar sensor to detect objects in an environment external to the PMVs. Specifically, the PMV may be equipped with a variety of sensors, such as a radar, a sonar sensor, a (optional) camera, an inertia measurement unit (IMU), and/or the like. The combination of a radar reflection signal and a sonar signal may provide measurements of characteristics such as a Doppler velocity and height information of a nearby object, which may be input to a machine learning classifier to determine the probability that the nearby object is a VRU. For another example, the reflection pattern from radar and ultrasonic may be used to input to a machine learning classifier to determine a type of the road surface, e.g., an asphalt road surface, a concrete sidewalk surface, a wet-grass lawn surface, and/or the like.