Abstract: In some embodiments, a computer-implemented method for managing resources of a fleet of unmanned aerial vehicles (UAVs) is provided. A computing system creates a mission record and one or more candidate records. Each candidate record of the one or more candidate records represents one or more resources for accomplishing a mission represented by the mission record. The computing system adds a mission node representing the mission record to a resource competition network graph (RCN graph). The computing system adds one or more candidate nodes representing the one or more candidate records to the RCN graph. The computing system determines an optimized allocation of candidate records to mission records using at least a subgraph of the RCN graph. A candidate record is determined to commit to a mission record, and the computing system updates the RCN graph to commit the candidate record to the mission record.

Abstract: An uncrewed aerial vehicle (UAV) may be configured to hover above a particular charging pad within a portion of a cluster of charging pads for UAVs. The cluster may include the charging pads arranged in a layout and fiducial markers distributed at positions across the layout. While hovering above the particular charging pad, the UAV may capture an aerial image of the portion of the cluster. The UAV may derive cluster-portion observation data from the image, the cluster-portion observation data including information indicating a position of the particular charging pad, and positions of one or more fiducial markers within the portion of the cluster relative to the particular charging pad. The UAV may send the cluster-portion observation data to a computing system in an infrastructure support network for UAVs, and thereafter receive, from the computing system, location information indicating that UAV's geolocation is a geolocation of the particular charging pad.

Abstract: A method includes charging a battery of a vehicle to a charge threshold voltage. The method also includes discharging the battery from the charge threshold voltage to a post-task voltage by performing a travel task using the vehicle. The method additionally includes determining that a battery calibration condition has been met. The method further includes, based on determining that the battery calibration condition has been met, discharging the battery from the post-task voltage to a discharge threshold voltage by performing a battery discharge task. The method yet further includes determining a capacity of the battery based on a first electrical output of the battery during the travel task and a second electrical output of the battery during the battery discharge task.

Abstract: A computer-implemented method includes obtaining an aerial image representing an object in an environment and providing the aerial image as input to a machine learning model. Based on the aerial image, and using the machine learning model, a textual description of a location of the object in the environment is generated and the textual description of the location of the object is outputted.

Abstract: In some embodiments, a method for autonomous navigation of an unmanned aerial vehicle (UAV) is provided. The UAV determines a tracked position using at least one positioning sensor of the UAV. The UAV captures an image using a camera of the UAV. The UAV determines a visual position confidence area using the captured image. The UAV checks the tracked position using the visual position confidence area to determine whether the tracked position is accurate. In response to determining that the tracked position is not accurate, the UAV causes a corrective action based on the visual position confidence area to be taken.

Abstract: A visual monitoring system for monitoring an autonomous flight operational area from a remote location is provided. The visual monitoring system can be configured to capture image data related to a verification target with one or more cameras, transmit the image data to a remote location, and display the image data related to the verification target for review by operational personnel at the remote location. The image data can be related to at least a first verification target having a first nominal condition. The operational personnel can review the image data and compare it to the data to the first nominal condition. If an off-nominal condition of the first verification target is detected by the operational personnel, the visual monitoring system can pause operations at the autonomous flight operational area until the off-nominal condition of the first verification target is resolved.

Abstract: A technique for a UAV includes acquiring a query aerial image with an onboard camera of the UAV and a reference aerial image, the query aerial image including multiple instances of an asset and the reference aerial image including annotated pixels indicating an expected location and an identification for the multiple instances of the asset. The technique further includes identifying a plurality of corresponding pixels between the query aerial image and the reference aerial image, determining a homography transformation describing a relationship between the query aerial image and the reference aerial image, annotating the query aerial image to identify a first instance of the asset included in the multiple instances of the asset within the query aerial image, and instructing the UAV to perform an action associated with the first instance of the asset.

Abstract: A method includes causing an aerial vehicle to deploy a tethered component to a particular distance beneath the aerial vehicle by releasing a tether connecting the tethered component to the aerial vehicle. The method also includes obtaining, from a camera connected to the aerial vehicle, image data that represents the tethered component while the tethered component is deployed to the particular distance beneath the aerial vehicle. The method additionally includes determining, based on the image data, a position of the tethered component within the image data. The method further includes determining, based on the position of the tethered component within the image data, a wind vector that represents a wind condition present in an environment of the aerial vehicle. The method yet further includes causing the aerial vehicle to perform an operation based on the wind vector.

Abstract: A technique for processing delivery aborts by a UAV delivery service includes: acquiring a delivery zone (DZ) image of a delivery destination including a DZ for a package being delivered to the delivery destination by a UAV; determining to abort a delivery mission for the package based upon the DZ image; converting the DZ image to a vector embedding; performing a similarity search on a vector database using the vector embedding, the vector database storing reference images of other delivery destinations indexed to reference vector embeddings and outcome attributes describing delivery outcomes associated with the reference images, and wherein the similarity search identifies a subset of the reference images deemed to have a threshold similarity to the DZ image; and prompting a vision language model with the DZ image and the subset of the reference images to provide an abort explanation or to determine a delivery disposition for the DZ.

Abstract: A technique for provisioning UAVs to operate in a specific geographical region includes acquiring reference aerial images of the specific geographical region from a secondary source. The reference aerial images have at least one style characteristic that is distinct from that of operational aerial images acquired by the UAVs during operation. The reference aerial images are transformed into training images that adopt a style of the operational aerial images using a neural style transfer. A neural network that facilitates vision-based navigation of the UAVs is trained using the training images. The neural network is trained to operate in the specific geographical region prior to operation in the specific geographical region.

Abstract: A method includes determining, by an unmanned aerial vehicle (UAV), a position of an autoloader device for the UAV; based on the determined position of the autoloader device, causing the UAV to follow a descent trajectory in which the UAV moves from a starting position to a first nudged position in order to deploy a tethered pickup component of the UAV to a payout position on an approach side of the autoloader device; deploying the tethered pickup component of the UAV to the payout position; causing the UAV to follow a side-step trajectory in which the UAV moves laterally to a second nudged position in order to cause the tethered pickup component of the UAV to engage the autoloader device; and retracting the tethered pickup component of the UAV to pick up a payload from the autoloader device.

Abstract: A method includes receiving an input specifying a starting location and a destination location for an aerial vehicle. The method additionally includes determining, based on the starting location and the destination location, an aerial path for the aerial vehicle to follow from the starting location to the destination location. The method also includes determining, based on the aerial path, a property of aerial image data, where the aerial image data is obtainable using the aerial vehicle while traversing the aerial path, and where the aerial image data represents an environment along the aerial path. The method further includes determining, based on the property, a path score associated with the aerial path, and outputting the aerial path based on the path score.

Abstract: A computer-implemented method comprises receiving an image captured by a camera on an unmanned aerial vehicle (UAV). The image depicts an environment below the UAV. A feature mask associated with the image is generated via a machine learning model that is trained to identify and semantically label pixels representing the environment depicted in the image. One or more reference tiles associated with the environment are retrieved. The reference tiles are associated with particular geographic locations and specify semantically labeled pixels representing the geographic locations. The semantically labeled pixels of the feature mask are correlated with the semantically labeled pixels of at least one of the one or more reference tiles to determine the geographic location of the UAV in the environment.

Abstract: A technique for using an onboard camera capable of day and night operation on an unmanned aerial vehicle (UAV) includes: traveling along a route of the UAV at night; and acquiring an aerial image with an onboard camera. The onboard camera includes: a sensor device for receiving photons and converting the photons into photoelectrons; a processor for processing the photoelectrons into an image file; a lens positioned adjacent to the sensor device for focusing the photons on the sensor device; and a filter positioned adjacent an outer surface of the lens. The filter can permit photons in the infrared light spectrum to pass to the lens, and attenuate at least a portion of the photons in the visible light spectrum prior to reaching the lens.

Abstract: A package coupling apparatus for securing a package to an unmanned aerial vehicle (UAV) is provided. The package coupling apparatus includes a support plate configured to be secured to an upper surface of the package and a handle extending up from the support plate. The handle includes a handle opening and a bridge that extends over the handle opening, wherein the bridge is configured to be secured by a component of the UAV.

Abstract: An uncrewed aerial vehicle (UAV) includes a fuselage extending along a first direction from a rear end to a front end. The fuselage has a cross-sectional area at an intermediate position between the front and rear end. An external payload storage area is positioned at the intermediate position and is configured to receive a payload that is secured to the fuselage and that extends laterally outward from the cross-sectional area of the fuselage. A drag reduction device is coupled to the fuselage. The drag reduction device has a length extending from a free end to an attached end that is secured to the fuselage, a width that extends perpendicular to the first direction of the fuselage, and a depth. In an operating position, the drag reduction device is positioned in front of the payload storage area, is spaced from the payload storage area, and extends outward from the fuselage.

Abstract: In some embodiments, a computer-implemented method for simulating an unmanned aerial vehicle (UAV) to improve control system performance is provided. A computing system obtains ground truth aerial imagery for a region that depicts the region during a first state. The computing system determines a route for a simulated UAV within the region. The computing system generates, based on the ground truth aerial imagery, predicted aerial imagery that depicts portions of the region associated with the route. The computing system generates simulated aerial imagery that depicts portions of the region associated with the route during a second state different from the first state by providing the predicted aerial imagery to a machine learning model. The computing system simulates travel of the simulated UAV along the route during the second state by providing the simulated aerial imagery as simulated input to at least one control system of the simulated UAV.

Abstract: In some embodiments, techniques are provided for analyzing time series data to detect anomalies. In some embodiments, the time series data is processed using a machine learning model. In some embodiments, the machine learning model is trained in an unsupervised manner on large amounts of previous time series data, thus allowing highly accurate models to be created from novel data. In some embodiments, training of the machine learning model alternates between a fitting optimization and a trimming optimization to allow large amounts of training data that includes untagged anomalous records to be processed. Because a machine learning model is used, anomalies can be detected within complex systems, including but not limited to autonomous vehicles such as unmanned aerial vehicles. When anomalies are detected, commands can be transmitted to the monitored system (such as an autonomous vehicle) to respond to the anomaly.

Abstract: A method for perception validation of an unmanned aerial vehicle (UAV) includes: acquiring an aerial image of a ground area with an onboard camera system of the UAV, generating a semantic above ground altitude (AGL) estimate with a neural network trained to output the semantic AGL estimate in response to the aerial image fed as an input to the neural network, generating a motion estimate or a position estimate based upon perception sensor data output from a perception sensor disposed onboard the UAV, and cross-validating the motion or position estimate against the semantic AGL estimate.

Abstract: A computing system in an infrastructure support network for uncrewed aerial vehicles (UAVs) may receive, from a UAV, aerial observation data of a ground-based cluster of charging pads for UAVs, the cluster comprising assets including the charging pads arranged in a layout and fiducial markers distributed across the layout. The aerial observation data may comprise position measurements of the UAV at aerial geolocations above the cluster, and vector positions of one or more assets with respect to the aerial geolocations. The computing system may generate a map graph from the aerial observation data, the map graph comprising (i) nodes corresponding to both the aerial geolocations and vector positions, and (ii) edges between pairs of selected nodes, the edges corresponding to distances between selected nodes and including measurement uncertainties. The computing system may generate a spatial map of cluster assets of the cluster by computationally optimizing the map graph.