Methods, systems and apparatus, including computer programs encoded on computer storage media for an unmanned aerial vehicle aerial survey. One of the methods includes receiving information specifying a location to be inspected by an unmanned aerial vehicle (UAV), the inspection including the UAV capturing images of the location. Information describing a boundary of the location to be inspected is obtained. Inspections to be assigned to the location are determined, with the inspection legs being parallel and separated by a particular width. A flight pattern is determined based on a minimum turning radius of the UAV, with the flight pattern specifying an order each inspection leg is to be navigated along, and a direction of the navigation.

A heading control system for an aircraft arranged to maintain a heading of an aircraft by controlling a nose wheel angle of the aircraft. The heading control system includes an interface arranged to receive a bias signal indicating a bias towards the port or the starboard of the aircraft and one or more processors. The one or more processors are arranged to determine, based on the bias signal, an offset angle defining an offset from a longitudinal axis of the aircraft and to perform a control process to control the nose wheel angle within an angular range based on the offset angle.

A heading control system for an aircraft arranged to maintain a heading of an aircraft by controlling a nose wheel angle of the aircraft. The heading control system includes an interface arranged to receive a bias signal indicating a bias towards the port or the starboard of the aircraft and one or more processors. The one or more processors are arranged to determine, based on the bias signal, an offset angle defining an offset from a longitudinal axis of the aircraft and to perform a control process to control the nose wheel angle within an angular range based on the offset angle.

Aerial navigation is disclosed. A system can detect a difference between forces applied to a plurality of ground contact points of an aerial vehicle taxiing on a ground surface. The system can determine an adjustment to a vertical component of a thrust. The thrust can be produced by at least one of a rotor or a propeller of the aerial vehicle to reduce the difference between the forces applied to the plurality of ground contact points of the aerial vehicle. The system can generate a control output to cause the at least one of the rotor or the propeller to adjust the vertical component of the thrust to reduce the difference between the forces.

Aerial navigation is disclosed. A system can detect a difference between forces applied to a plurality of ground contact points of an aerial vehicle taxiing on a ground surface. The system can determine an adjustment to a vertical component of a thrust. The thrust can be produced by at least one of a rotor or a propeller of the aerial vehicle to reduce the difference between the forces applied to the plurality of ground contact points of the aerial vehicle. The system can generate a control output to cause the at least one of the rotor or the propeller to adjust the vertical component of the thrust to reduce the difference between the forces.

In various examples, live perception from sensors of a vehicle may be leveraged to generate potential paths for the vehicle to navigate an intersection in real-time or near real-time. For example, a deep neural network (DNN) may be trained to compute various outputssuch as heat maps corresponding to key points associated with the intersection, vector fields corresponding to directionality, heading, and offsets with respect to lanes, intensity maps corresponding to widths of lanes, and/or classifications corresponding to line segments of the intersection. The outputs may be decoded and/or otherwise post-processed to reconstruct an intersectionor key points corresponding theretoand to determine proposed or potential paths for navigating the vehicle through the intersection.

In various examples, live perception from sensors of a vehicle may be leveraged to generate potential paths for the vehicle to navigate an intersection in real-time or near real-time. For example, a deep neural network (DNN) may be trained to compute various outputssuch as heat maps corresponding to key points associated with the intersection, vector fields corresponding to directionality, heading, and offsets with respect to lanes, intensity maps corresponding to widths of lanes, and/or classifications corresponding to line segments of the intersection. The outputs may be decoded and/or otherwise post-processed to reconstruct an intersectionor key points corresponding theretoand to determine proposed or potential paths for navigating the vehicle through the intersection.

An object classification system may be present on a vehicle with a computing device connected to a sensor mounted on the vehicle. In response to an object positioned in a field of view of the sensor, the computing device may determine a state of the object after calculating a relative shape stability value for the object over time. The computing device may then deviate from a predetermined route for the vehicle in response to a classification of the object as a dynamic state.

An aircraft tow vehicle comprises a turntable lifting unit configured to automatically rotate and lift for attachment to a nose landing gear of an aircraft. A gate coupled to the turntable lifting unit automatically unlocks, opens to receive the nose landing gear, closes to secure the nose landing gear, and locks. A sensor system detects the nose landing gear. A controller receives data from the sensor system, processes the data to determine a position of the nose landing gear, and controls the turntable lifting unit while automatically adjusting a position of the tow vehicle relative to the nose landing gear. A moving floor adjusts to accommodate different nose wheel sizes. A nose wheel adapter automatically positions itself to hold down the nose landing gear when weight is detected on the moving floor.

An aircraft tow vehicle comprises a turntable lifting unit configured to automatically rotate and lift for attachment to a nose landing gear of an aircraft. A gate coupled to the turntable lifting unit automatically unlocks, opens to receive the nose landing gear, closes to secure the nose landing gear, and locks. A sensor system detects the nose landing gear. A controller receives data from the sensor system, processes the data to determine a position of the nose landing gear, and controls the turntable lifting unit while automatically adjusting a position of the tow vehicle relative to the nose landing gear. A moving floor adjusts to accommodate different nose wheel sizes. A nose wheel adapter automatically positions itself to hold down the nose landing gear when weight is detected on the moving floor.