Autonomous Drone Navigation Techniques and Improvements

A path planning system for unmanned aerial vehicles (UAVs) that considers local wind and turbulence conditions estimates wind/turbulence distributions at a given altitude and generates a cost map. This cost map is used to calculate flight paths that avoid areas with high wind/turbulence during path planning.

Source

An unmanned aerial vehicle (UAV) that can autonomously fly, detect objects, avoid collisions, and land safely. The UAV has sensors like a camera and radar to detect objects, and a controller that analyzes the sensor data to determine if an override condition exists. If so, it enters a hover mode where it hovers in place for remote control.

Source

Displaying safe, collision-free paths for drones using 3D point clouds. The method involves displaying a 3D airspace using point clouds containing vector points. Paths are generated and displayed in this 3D airspace by connecting selected vector points with corridors. The corridor widths are determined by the vector point sizes. The paths can be verified and simulated for safety. Obstacles can be detected and avoided by deflecting paths to adjacent vector points or using Bezier curves.

Source

Safely determining the flight path of an unmanned aerial vehicle using multiple position determination systems to provide redundancy and error checking. The method involves using a primary position system like GPS along with a secondary position system like image-based or lidar to independently determine the vehicle's position. Then a plausibility check is performed comparing the two sets of position data. If they pass the check, the primary position is used. This allows reliable flight path determination even if the primary system fails or has errors.

Source

Detecting and alleviating the effects of optical discrepancies in images used to guide autonomous navigation by a vehicle such as an unmanned aerial vehicle (UAV). The method involves detecting optical discrepancies caused by issues like dirt on the camera lens. The discrepancies are detected by tracking photometric differences between corresponding pixels in stereo images over time. The discrepancies are used to generate an image mask that ignores error-prone regions. The discrepancies can also trigger cleaning the lens or notifying the user.

Source

Using machine learning to guide unmanned aerial vehicles (UAVs) and other platforms around obstacles without expensive imaging systems. The approach involves training ML models to generate guidance information like object locations based on time-of-arrival (TOA) data from sensors. This avoids the computational expense of processing images to identify obstacles in real time on board the platform.

Source

Monitoring and managing drone traffic in shared airspace for safe and efficient navigation. The system plots navigation paths for drones to move through a controlled space based on monitoring existing drones and known obstacles. When a new drone requests to enter the space, it provides them a safe path through the monitored airspace. The paths consider weather, restricted airspace, drone charge levels, and traffic. The paths can also be encrypted or secured with blockchain.

Source

Aircraft control system that uses a fade function to predict aircraft paths and avoid terrain obstacles. The system takes pilot input commands and attenuates them over time using a fade function. These attenuated commands are then used to predict aircraft paths. If any of the predicted paths intersect with terrain obstacles, the system generates new commands to avoid the obstacle.

Source

Use of autonomous vehicles like UAVs for capturing images with intuitive high-level control objectives that abstract away the complexity of autonomous flight. Developers can use an API to specify objectives like tracking objects, capturing interesting scenes, etc. These objectives are used by the autonomous navigation system for flight planning. Developers can also build skills that modify objectives and control the vehicle behavior during flight. The skills can be shared, learned from, and used for visual outputs like tracking and recognition to enhance autonomous performance.

Source

A flight control system for unmanned aerial vehicles that enables the safe passing of nearby aircraft. The system detects nearby aircraft and determines if passing is possible based on their trajectories and airspace conditions. If passing is possible, it controls the drone to perform a passing maneuver at a safe distance from the other aircraft.

Source

Navigation system for unmanned aerial vehicles (UAVs) that does not rely on GPS and can work in urban environments with limited satellite coverage. The system uses two synchronized periodic wideband signals transmitted from base stations. The UAV receives the signals and determines its position relative to the base stations based on the signal reception times and intensities. This allows the UAV to navigate along a flight path defined by the base stations without GPS.

Source

Unmanned aerial vehicle (UAV) orbiting method and device using computer vision instead of GPS to autonomously circle around and film an object. A binocular camera on the UAV captures a desired object and orbit parameters. The UAV detects the camera's position, orientation, and distance from the object in real-time. It then autonomously orbits the object using those parameters, allowing it to circle and film objects without GPS dependency and with consistent framing.

Source

Controlling an unmanned aerial vehicle (UAV) to automatically assist in navigating over sloping terrain. The technique involves using the UAV's sensors to measure distances to the ground and determine the slope angle. The flight control system then adjusts the vertical and horizontal thrust to move the UAV along the slope. This allows the UAV to autonomously climb or descend slopes without requiring precise manual control inputs.

Source

Autonomous vehicle (AV) team coordination system that allows AVs to continue functioning as a team even when communication is disrupted. The system uses controllers on each AV to store mission data including objectives and default trajectories. If communication is lost, AVs can estimate teammate trajectories based on the stored mission data.

Source

4D path planning for unmanned vehicles like drones using point cloud data to define safe and efficient flight paths. The method involves creating flight corridors in a 3D airspace by using point cloud information. The corridors are verified and simulated to ensure they are collision-free. The resulting paths are displayed and stored for unmanned vehicle navigation.

Source

Deploying a visual navigation system for UAVs that provides reliable positioning when GPS signals are unavailable or unreliable. The technique involves placing multiple geo-fiducials around a landing pad, each with a unique offset and direction from a surveyed center point. The UAV uses computer vision to recognize and triangulate the geo-fiducials for precise navigation.

Source

Using autonomous navigation to capture images of objects in the environment identified by a user. The system fuses inputs from onboard sensors and other devices to navigate the aerial vehicle, locate the object, and maneuver the vehicle for capturing images. It leverages techniques like perception from mobile devices, 3D mapping, object tracking, and deep learning to enable user-directed image capture by the autonomous aerial vehicle.

Source

A flight controller using a state machine that provides a high level of abstraction for controlling autonomous drones and vehicles. The state machine has states like initialization, arming, takeoff, mission, landing, and teleoperation. The controller transitions between states based on vehicle conditions and user commands. The state machine interacts with lower level modules for functions like obstacle avoidance. It enables fault detection and recovery, remote control override, and coordinated multi-vehicle missions.

Source

Modifying flight of an unmanned aerial vehicle (UAV) without taking manual control of the vehicle. The UAV has a unique interface that allows the user to give general instructions to the autonomous flight system, and also to modify the flight path. This provides an intuitive way to interact with the autonomous UAV without overriding or disrupting the autonomous operation. The system receives a first user input to initiate autonomous flight. Then it receives a second user input to modify the autonomous flight. The UAV's flight controller generates signals to effect the autonomous flight from the first input, and separate signals to modify the flight from the second input. This allows the user to directly affect the UAV's flight path and direction during autonomous operation while still achieving the mission goal.

Source

Enabling an autonomous vehicle like a drone to safely navigate and inspect objects using a combination of flight planning, model optimization, and real-time sensing. The navigation system plans an initial inspection path based on a simplified model of the object. During flight, sensors provide detailed object data that is used to refine the model and update the inspection path. This iterative process allows for improving object models and paths in flight to avoid obstacles and optimize inspection coverage. It also uses pose optimization to align sensor data with the model.

Source