Neural network-guided passive sensor drone inspection system
AssigneeDrobotics, LLC
InventorEyal Stein
Life of this patent
Estimating the term…
Abstract
A drone system for collecting structural condition data about a structure having an array of sensors disposed at various locations on the structure and methods of using such a drone system are disclosed herein. The drone inspection system leverages neural networks to calculate a drone flight path to classify the location of passive sensors and calculate a drone flight path to collect structural condition data about the structure using line of sight sensors for digital twin generation. Some of the sensors disposed on the structure may be passive sensors that comprise energy harvesters and must be energized to report the structural collection data to the drone. The drone inspection system may comprise an energy transfer module for energizing the passive sensor via the energy harvester.
Claims31
An unmanned aerial vehicle (UAV) system for collecting structural condition data of a structure having an array of passive sensors at various locations on the structure, each passive sensor energized by an energy harvester, the system comprising:
- a UAV navigation system;
- a UAV image acquisition device;
- a UAV agile transceiver array;
- a computer memory;
- a drone inspection neural network including a neural compute engine;
- a UAV energy transfer module targetable to energize an energy harvester such that when energized, the energy harvester energizes a one of the passive sensor;
- and
- a UAV communications device for receiving data from an energized sensor,
- wherein the neural compute engine is configured to use real time data from one or more of the UAV image acquisition device, the UAV agile transceiver array, or the UAV navigation system as inputs to a neural network computationally trained to
- (a) classify a passive sensor or energy harvester on an asset,
- (b) calculate a revised flight path for the UAV system to create a line of sight between the UAV energy transfer module and the energy harvester,
- (c) aim the energy transfer module at the energy harvester,
- (d) detect anomalies in UAV asset inspection data and passive sensor data in real time;
- and
- wherein the neural compute engine is further configured to request and receive neural network weight and configuration file updates in real time in response to anomalies detected in UAV asset inspection data and passive sensor data in real time, classify new anomalies that do not have a database record, and to calculate the revised flight path for further inspection of the asset and anomaly.
A method for collecting structural condition data of a structure having an array of passive sensors at various locations on the structure, each passive sensor energized by an energy harvester, the method comprising the steps of:
- providing an unmanned aerial vehicle (UAV) having a UAV navigation system, a UAV image acquisition device, a UAV agile transceiver array, a computer memory, and a drone inspection neural network including a neural compute engine, wherein the neural compute engine is configured to request and receive neural network weight and configuration file updates in real time in response to anomalies detected in UAV asset inspection data and passive sensor data in real time, classify new anomalies that do not have a database record, and to calculate a revised flight path for further inspection of the asset and anomaly;
- providing a UAV energy transfer module targetable to energize an energy harvester;
- providing a UAV communications device for receiving data from an energized sensor;
- obtaining real time data from one or more of the UAV image acquisition device, the UAV agile transceiver array, or the UAV navigation system;
- using the real time data as inputs to the drone inspection neural network to classify a passive sensor or energy harvester on an asset;
- using the real time data as inputs to the drone inspection neural network to calculate the revised flight path for the UAV system to create a line of sight between the UAV energy transfer module and the energy harvester;
- using the real time data as inputs to the drone inspection neural network to aim the energy transfer module at the energy harvester;
- and
- using the real time data as inputs to the drone inspection neural network to detect anomalies in UAV asset inspection data and passive sensor data in real time.
Dependencies are inferred from the claim text.