Drone Flight Control Using Virtual Rails for Collision Avoidance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing drone controllers fail to ensure safe flight in crowded areas, leading to collisions between drones and objects, and between multiple drones flying simultaneously for cinematography, as they lack effective collision avoidance and field-of-view management.
Innovation Solution
A drone system with a controller that uses predefined virtual rails and a mathematical model to predict and minimize errors, applying control inputs to actuators to avoid collisions and optimize flight paths, while considering constraints such as collision avoidance and camera field-of-view.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing controllers are used to control drone flight, then the drone can fly and capture images, but the drone cannot avoid collision with subjects or other drones
Solution Approach 1:
The controller pre-calculates multiple future control inputs (control sequence) based on the cost function and mathematical model before executing any single control action. This preliminary planning allows the drone to anticipate and avoid collisions with subjects or other drones by evaluating future states, rather than reacting to current conditions alone.
Solution Approach 2:
The system dynamically adjusts control inputs by continuously minimizing the cost function that incorporates both tracking performance and collision avoidance. The mathematical model of the drone enables real-time prediction of future states, allowing the controller to adapt control actions dynamically to prevent collisions while maintaining flight stability.
2Productivity
If multiple drones are flown simultaneously for cinematography, then diverse camera views are captured, but drones collide with each other or enter each other's field of view
Solution Approach 1:
The cost function incorporates feedback from the mathematical model to evaluate how current control actions affect future drone positions and camera field of view. By continuously minimizing this cost function, the controller receives feedback on potential inter-drone interference and adjusts control inputs to maintain safe separation and optimal filming positions.
Solution Approach 2:
The controller pre-calculates control sequences that anticipate future drone positions and potential field-of-view conflicts before they occur. This preliminary action allows multiple drones to coordinate their movements proactively, ensuring they capture diverse views without colliding or interfering with each other's camera coverage.
3Adaptability or versatility
If the drone follows a predefined path, then the flight path is controlled and predictable, but the drone cannot adapt to avoid moving obstacles or dynamic conditions
Solution Approach 1:
The controller pre-calculates a sequence of control inputs based on the current state and cost function, creating a planned adaptation path before executing any single control action. This preliminary planning allows the drone to adapt to dynamic conditions like moving obstacles while maintaining smooth, predictable flight characteristics.
Solution Approach 2:
The system dynamically adapts the flight path by continuously minimizing the cost function that balances path following with obstacle avoidance. The mathematical model enables real-time prediction of how control actions will affect drone trajectory, allowing adaptive path adjustment while maintaining flight stability and control simplicity.
Data Source
AI summary
According to the present invention there is provided a drone (1) comprising one or more propellers (2) and one or more actuators (3) for actuating said one or more propellers (2) to generating a thrust force which enables the drone (1) to fly; a controller (4) which is configured such that it can control the flight of the drone (1), wherein the controller (4) comprises a memory (6) having stored therein a plurality of predefined sets of positions which define a virtual rail which can be used to guide the flight of the drone (1) so that the drone can avoid collision with an subject; and wherein the controller further comprises a mathematical model (7) of the drone; wherein the controller (4) is configured to control the flight of the drone by performing at least the following steps, (a) approximating lag error based on the position of the drone (1) measured by a sensor (5) and the virtual rail, wherein the lag error is the distance between a point along the virtual rail which is closest to the drone (1) and an estimate of said point along the virtual rail which is closest to the drone (1); (b) approximating a contour error based on the position of the drone (1) as measured by a sensor (5) and the virtual rail, wherein the contour error is the distance between a point along the virtual rail which is closest to the drone (1) and the position of the drone (1); (c) defining a cost function which comprises at least said approximation of the lag error and said approximation of the contour error; (d) minimizing the defined cost function, while also respecting at least limitations of the drone which are defined in said mathematical model, to determine a plurality of control inputs over a predefined time period into the future, and (e) applying the first control input only to the one or more actuators (3). There is further provided a corresponding method for controlling the flight of a drone.