Object tracking by an unmanned aerial vehicle using visual sensors

Patent metadata

Field	Value
Publication number	US-11295458-B2
Application number	US-201715827945-A
Country	US
Kind code	B2
Filing date	Nov 30, 2017
Priority date	Dec 1, 2016
Publication date	Apr 5, 2022
Grant date	Apr 5, 2022

Abstract

Systems and methods are disclosed for tracking objects in a physical environment using visual sensors onboard an autonomous unmanned aerial vehicle (UAV). In certain embodiments, images of the physical environment captured by the onboard visual sensors are processed to extract semantic information about detected objects. Processing of the captured images may involve applying machine learning techniques such as a deep convolutional neural network to extract semantic cues regarding objects detected in the images. The object tracking can be utilized, for example, to facilitate autonomous navigation by the UAV or to generate and display augmentative information regarding tracked objects to users.

First claim

What is claimed is:

1. A method for tracking physical objects in a physical environment, the method comprising:
receiving, by a computer system of an autonomous vehicle, images of the physical environment captured by one or more image capture devices coupled to the autonomous vehicle;
processing, by the computer system, the received images to:
detect physical objects in the physical environment associated with a particular class of physical objects,
distinguish one or more instances of the physical objects from a background of the received images, and
extract semantic information regarding the detected one or more physical objects;

predict, based on the received images, the extracted semantic information and a motion model associated with the particular class of physical objects, a trajectory of a particular physical object instance of the one or more instances of the physical objects through three-dimensional (3D) space of the physical environment;
tracking, by the computer system, a 3D trajectory of the particular physical object instance of the one or more instances of the physical objects through the 3D space of the physical environment based, at least in part, on the predicted trajectory of the particular physical object instance;
generating and continually updating, by the computer system, based on the tracked 3D trajectory of the particular physical object instance, a planned 3D trajectory for the autonomous vehicle through the physical environment that follows the tracked 3D trajectory of the particular physical object instance; and
generating, by the computer system, control commands configured to cause the autonomous vehicle to maneuver along the planned 3D trajectory.

2. The method of claim 1 , wherein the received images are processed using a deep convolutional neural network.

3. The method of claim 1 , wherein processing the received images to detect the one or more physical objects associated with the particular class of physical objects includes:
generating a dense per-pixel segmentation based on the received images, wherein each pixel in the dense per-pixel segmentation is associated with a value indicative of a likelihood that the pixel corresponds with the particular class of physical objects.

4. The method of claim 3 , the dense per-pixel segmentation is one of a plurality of dense per-pixel segmentations comprising a tensor, each of the plurality of dense per-pixel segmentations associated with a different class of physical objects.

5. The method of claim 3 , wherein processing the received images to distinguish one or more instances of the detected one or more physical objects includes:
analyzing the dense per-pixel segmentation generated based on the received images to associate pixels corresponding to the particular class of physical objects with a particular instance of the particular class of physical objects.

6. The method of claim 5 , wherein associating pixels corresponding to the particular class of physical objects with the particular instance of the particular class of physical objects includes:
applying a grouping process to group:
pixels that are substantially similar to other pixels associated with the particular instance;
pixels that are spatially clustered with other pixels associated with the particular instance; and/or
pixels that fit an appearance-based model for the particular class of physical objects.

7. The method of claim 1 , wherein the semantic information includes information regarding any of a position, orientation, shape, size, scale, appearance, pixel segmentation, or activity of the detected one or more physical objects.

8. The method of claim 1 , further comprising:
receiving, by the computer system, sensor data from one or more other sensors coupled to the autonomous vehicle; and
processing, by the computer system, the received sensor data with the received images using a spatiotemporal factor graph to predict a 3D trajectory of the particular physical object instance;
wherein the tracking is based on the predicted 3D trajectory of the particular physical object instance.

9. The method of claim 1 , further comprising:
generating, by the computer system, control commands configured to cause a gimbal mechanism to adjust an orientation of the image capture device relative to the autonomous vehicle so as to keep the tracked particular physical object instance within a field of view of the image capture device.

10. The method of claim 1 , further comprising:
generating, by the computer system, an augmentation based on the tracking of the particular physical object instance; and
causing, by the computer system, the generated augmentation to be presented at an augmented reality (AR) device.

11. The method of claim 1 , further comprising:
tracking, by the computer system, a second 3D trajectory of a second particular physical object instance while tracking the particular physical object instance; and
generating, by the computer system, an output based on the tracking second tracked 3D trajectory of the second particular physical object instance.

12. The method of claim 1 , wherein the autonomous vehicle is an unmanned aerial vehicle (UAV).

13. The method of claim 1 , wherein the particular class of physical objects is selected from a list of classes of physical objects comprising people, animal, vehicles, buildings, landscape features, and plants.

14. An unmanned aerial vehicle (UAV) configured for autonomous flight through a physical environment, the UAV comprising:
a first image capture device;
a second image capture device; and a computer system configured to:
receive images of the physical environment captured by any of the first image capture device or second image capture device;
process the received images to detect one or more physical objects in the physical environment associated with a particular class of physical objects;
identify a motion model associated with the particular class of physical objects;
process the received images to distinguish one or more instances of the detected one or more physical objects;
process the received images to extract semantic information regarding the detected one or more physical objects;
track a three-dimensional (3D) trajectory of a particular physical object instance of the detected one or more physical objects based on the received images, the extracted semantic information and a motion model associated with the particular class of physical objects;
generate and continually update, based on the tracked 3D trajectory of the particular physical object instance, a planned 3D trajectory for the UAV through the physical environment that follows the tracked 3D trajectory of the particular physical object instance; and
generate control commands configured to cause the UAV to maneuver along the planned 3D trajectory so as to cause the UAV to follow the particular physical object instance through the physical environment in real time.

15. The UAV of claim 14 , wherein the received images are processed using a deep convolutional neural network.

16. The UAV of claim 14 , wherein processing the received images to detect the one or more physical objects associated with the particular class of physical objects includes:
generating a dense per-pixel segmentation of the received image, wherein each pixel in the received image is associated with a value indicative of a likelihood that the pixel corresponds with the particular class of physical objects.

17. The UAV of claim 14 , wherein processing the received images to distinguish one or more instanc

Assignees

• Skydio Inc

Inventors

• Dasgupta Saumitro
• Martirosyan Hayk
• Koppula Hema
• Kendall Alex
• Stone Austin
• Donahoe Matthew
• Bachrach Abraham Galton
• Bry Adam Parker

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06V20/17

  taken from planes or by drones · CPC title

• G06V10/82

  using neural networks · CPC title

• G06V20/13 Primary

  Satellite images · CPC title

• G06T7/292 Primary

  Multi-camera tracking · CPC title