Monocular Drone Obstacle Detection Using Planned Parallax Motion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing obstacle detection systems for drones, particularly at low altitudes, struggle to accurately detect objects below the horizon using monocular imaging due to reliance on traditional methods that require additional sensors or multiple cameras, leading to increased complexity and cost.
Innovation Solution
A method and apparatus for monocular obstacle detection on a drone flight path using a single camera that performs planned perpendicular motions to create a stereo effect, allowing depth estimation and obstacle detection without additional hardware, adjusting motion parameters based on confidence levels and threshold comparisons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional obstacle detection methods are used with monocular imaging, then obstacle detection can be performed, but additional sensors or multiple cameras are required, increasing device complexity and cost
Solution Approach 1:
The patent combines the functions of multiple cameras into a single monocular camera by using temporal sequencing. The monocular camera captures first and second frames at different time points, effectively merging the stereo vision capability of multiple cameras into one camera through time-based separation of viewing angles.
Solution Approach 2:
The patent transitions from spatial separation (multiple cameras at different positions) to temporal separation (single camera at different time points). By instructing the drone to perform planned motion between capturing the first and second frames, the system creates a stereo effect along the time dimension rather than the space dimension, achieving depth estimation with a single camera.
2Measurement precision
If multiple cameras are used for obstacle detection, then depth estimation accuracy improves, but hardware cost and system complexity increase
Solution Approach 1:
The system uses periodic motion of the drone to create alternating viewing angles. The drone is instructed to perform planned motion that periodically changes its position relative to the flight path, allowing the monocular camera to capture images from different perspectives at regular intervals, thereby enabling depth estimation without multiple cameras.
Solution Approach 2:
The patent creates a virtual stereo pair by capturing two images from a single camera at different time points. The first frame and second frame serve as copies of the scene from slightly different viewpoints, replicating the effect of having two physical cameras without the associated hardware complexity.
3Device complexity
If a monocular camera is used without planned motion, then hardware simplicity is maintained, but obstacle detection accuracy below the horizon deteriorates
Solution Approach 1:
The system transitions from a static camera setup to a dynamic one by incorporating planned motion. The drone is instructed to perform specific motion patterns between capturing the first and second frames, making the camera system dynamic rather than static. This dynamic approach enables the monocular camera to gather depth information that would otherwise require multiple fixed cameras.
Solution Approach 2:
The system performs preliminary motion planning before obstacle detection. The drone is instructed to execute planned motion that positions it optimally for capturing stereo-like images. This preliminary action of moving to specific positions enables the monocular camera to achieve accurate obstacle detection below the horizon without requiring complex hardware.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate obstacle detection without additional sensors, reducing production and deployment costs, conserving flight resources, and improving robustness under challenging conditions by dynamically adjusting stereo sensitivity and resolution.
Implementation Method 1
A monocular camera mounted on the drone is operated to capture at least one pair of images, each pair comprising a first frame captured along the flight path and a second frame captured at a deviation from the flight path
Data Source
AI summary
A method, apparatus, and computer program product for monocular obstacle detection on a drone flight path. A drone is instructed to move forward at a constant horizontal speed and concurrently perform a planned motion in at least one perpendicular direction. A monocular camera mounted on the drone is operated to capture one or more image pairs comprising at least a first frame captured along the flight path and a second frame captured at a deviation from the flight path effected by the planned motion. The first and second frames are analyzed to determine stereo matching and according thereto estimate a motion vector between the first and second frames for at least one region of interest. In response to determining according to the motion vector that an obstacle is present on the flight path, at least one obstacle detection response pattern is applied.