Systems and methods of sensor data fusion
Systems and methods of sensor data fusion including sensor data capture, curation, linking, fusion, inference, and validation. The systems and methods described herein reduce computational demand and processing time by curating data and calculating conditional entropy. The system is operable to fuse data from a plurality of sensor types. A computer processor optionally stores fused sensor data that the system validates above a mathematical threshold.
1 . A system for sensor data fusion for sensor management and utilization in autonomous transportation, comprising:
at least one computer processor including a memory;
at least one curation engine, at least one link engine, at least one fusion engine, at least one inference engine, and at least one validation engine;
at least one first distance sensor operable to capture a first distance measurement data from a vehicle to at least one object; and
at least one second distance sensor operable to capture a second distance measurement data from the vehicle to the at least one object;
wherein the at least one computer processor is operable to analyze the first distance measurement data and the second distance measurement data;
wherein the at least one curation engine is operable to curate the first distance measurement data and the second distance measurement data, creating a curated first distance measurement data and a curated second distance measurement data, the at least one link engine is operable to link the curated first distance measurement data and the curated second distance measurement data, the at least one fusion engine is operable to fuse the curated first distance measurement data and the curated second distance measurement data, creating fused data, the at least one inference engine is operable to determine at least one inference from the fused data, and the at least one validation engine is operable to validate the fused data;
wherein the at least one link engine is operable to link the curated first distance measurement data and the curated second distance measurement data based in part on at least one first distance property and at least one second distance property creating a property link;
wherein the at least one link engine is operable to calculate a conditional property entropy between the at least one first distance property and the at least one second distance property;
wherein the property link is a non-existent link, a weak link, or a strong link via the conditional entropy; and
wherein the at least one computer processor is operable to instruct a brake controller of the vehicle to brake based on the fused data.
2 . The system of claim 1 , wherein the at least one curation engine is operable to use artificial intelligence to determine the at least one first distance property and the at least one second distance property for the at least one link engine to calculate the conditional property entropy.
3 . The system of claim 1 , wherein the non-existent link is a conditional property entropy of 1, the weak link is a conditional property entropy of between less than 1 and greater than or equal to 0.33, and the strong link is a conditional property entropy of between less than 0.33 and 0.
4 . The system of claim 1 , wherein the at least one fusion engine creates at least one new data set.
5 . The system of claim 1 , wherein the at least one link engine is operable to iteratively calculate the conditional property entropy of the weak link until the at least one link engine calculates the non-existent link or the strong link.
6 . The system of claim 1 , wherein the at least one link engine is operable to link the curated first distance measurement data and the curated second distance measurement data in real-time.
7 . The system of claim 1 , wherein the system does not store the first distance measurement data and/or the second distance measurement data unless the at least one validation engine validates fused sensor data.
8 . The system of claim 1 , wherein the at least one first distance sensor and/or the at least one second distance sensor is operable to include a Light Detection and Ranging (LiDAR) sensor, a radar sensor, an ultrasonic sensor, a visible spectrum camera, a Global Positioning System (GPS) sensor, an infrared sensor, and/or a depth camera.
9 . A method for sensor data fusion for sensor management and utilization in autonomous transportation, comprising:
providing at least one computer processor including a memory;
providing at least one curation engine, at least one link engine, at least one fusion engine, at least one inference engine, and at least one validation engine;
at least one first distance sensor capturing a first distance measurement data from a vehicle to at least one object;
at least one second distance sensor capturing a second distance measurement data from the vehicle to the at least one object;
analyzing by the at least one computer processor the first distance measurement data and the second distance measurement data;
curating by the at least one curation engine the first distance measurement data and the second distance measurement data, creating a curated first distance measurement data and a curated second distance measurement data, linking by the at least one link engine the curated first distance measurement data and the curated second distance measurement data, fusing by the at least one fusion engine the curated first distance measurement data and the curated second distance measurement data, creating fused data, determining by the at least one inference engine at least one inference from the fused data, and validating by the at least one validation engine the fused data;
linking by the at least one link engine the curated first distance measurement data and the curated second distance measurement data based in part on at least one first distance property and at least one second distance property creating a property link;
calculating by the at least one link engine a conditional property entropy between the at least one first distance property and the at least one second distance property;
wherein the property link is a non-existent link, a weak link, or a strong link via the conditional property entropy; and
instructing by the at least one computer processor a brake controller of the vehicle to brake based on the fused data.
10 . The method of claim 9 , wherein the non-existent link is a conditional property entropy of 1, the weak link is a conditional property entropy of between less than 1 and greater than or equal to 0.33, and the strong link is a conditional property entropy of between less than 0.33 and 0.
11 . The method of claim 9 , further comprising categorizing via the at least one curation engine the at least one first distance property and the at least one second distance property for the at least one link engine to calculate the conditional property entropy based in part on historical accuracy of previous categorizations using artificial intelligence.
12 . The method of claim 9 , further comprising the at least one fusion engine creating at least one new data set.
13 . The method of claim 9 , further comprising iterating calculations so that the at least one link engine continuously calculates the conditional property entropy of the weak link until the at least one link engine calculates the non-existent link or the strong link.
14 . The method of claim 9 , further comprising dynamically adjusting via the at least one link engine a threshold for the non-existent link, the weak link, or the strong link.
15 . A system for sensor data fusion for sensor management and utilization in autonomous transportation, comprising:
at least one computer processor including a memory;
at least one curation engine, at least one link engine, at least one fusion engine, at least one inference engine, and at least one validation engine; and
at least two sensors, each of the at least two sensors operable to measure a first distance from a vehicle to at least one object and a second distance from the vehicle to the at least one object;
wherein the at least one computer processor is operable to analyze the first distance and the second distance;
wherein the at least one curation engine is operable to curate the first distance and the second distance, creating a curated first distance and a curated second distance, the at least one link engine is operable to link the curated first distance and the curated second distance, the at least one fusion engine is operable to fuse the curated first distance and the curated second distance, creating fused data, the at least one inference engine is operable to determine at least one inference from the fused data, and the at least one validation engine is operable to validate the fused data;
wherein the at least one link engine is operable to link the curated first distance and the curated second distance based in part on at least one first distance property and at least one second distance property creating a property link;
wherein the at least one first distance property is the same as the at least one second distance property;
wherein the at least one link engine is operable to calculate a conditional property entropy between the at least one first distance property and the at least one second distance property;
wherein the property link is a non-existent link, a weak link, or a strong link via the conditional property entropy; and
wherein the at least one computer processor is operable to instruct a brake controller of the vehicle to brake based on the fused data.
16 . The system of claim 15 , wherein the non-existent link is a conditional property entropy of 1, the weak link is a conditional property entropy of between less than 1 and greater than or equal to 0.33, and the strong link is a conditional property entropy of between less than 0.33 and 0.
17 . The system of claim 15 , wherein the at least one link engine is operable to link the curated first distance and the curated second distance in real-time.
18 . The system of claim 15 , wherein the at least one fusion engine creates at least one new data set.
19 . The system of claim 15 , wherein the system does not store the first distance or the second distance unless the at least one validation engine validates fused sensor data.
20 . The system of claim 15 , wherein the at least one link engine is operable to dynamically adjust a threshold for the non-existent link, the weak link, or the strong link.