Utilizing a plurality of on-road vehicles for increasing descriptive accuracy of areas and objects
A plurality of on-road vehicles moving along streets of a city, while utilizing onboard cameras for autonomous driving functions, are bound to capture, over time, random images of various city objects such as pedestrians, vehicles, building and other structures. Images of a certain city object may be captured by many of the vehicles at various times while passing by that object, thereby resulting in a corpus of imagery data containing various random images and other representations of that certain object. Per each of at least some of the city objects appearing more than once in the corpus of imagery data, appearances are detected and linked, and then analyzed and processed to result in representations and other data sets and types having a better descriptive accuracy in conjunction with the city object. The data sets having a better descriptive accuracy may then be used to reach certain conclusions or facilitate actions.
1 . A system operative to utilize a plurality of on-road vehicles for increasing descriptive accuracy of areas and/or objects, comprising:
a plurality of data interfaces located respectively onboard a plurality of on-road vehicles moving in a certain geographical area;
a plurality of data gathering configurations located respectively onboard said plurality of on-road vehicles and associated respectively with said plurality of data interfaces; and
a server;
wherein:
each of the data interfaces is configured to: (i) use the respective data gathering configuration to create a representation of areas and/or objects surrounding locations visited by the respective on-road vehicle, in which said representation comprises an inherent inaccuracy, and (ii) send said representation to the sever;
the server is configured to receive said plurality of representations respectively from the plurality of on-road vehicles, in which the plurality of representations comprises respectively the plurality of inherent inaccuracies; and
the server is further configured to correlate, in a post-capture analysis, from the plurality of representations received from the plurality of on-road vehicles, at least two representations of a same area and/or object, wherein said at least two representations were captured by different ones of said plurality of on-road vehicles at different times; and process said correlated at least two representations into a fused representation using at least one data combining technique, in which said fused representation comprises a better descriptive accuracy.
2 . The system of claim 1 , wherein the plurality of data gathering configurations are a plurality of lidar (light-imaging-detection-and-ranging) sensors.
3 . The system of claim 2 , wherein said inherent inaccuracy is associated with an inherent inaccuracy of a laser beam angle and/or timing in the lidar sensor.
4 . The system of claim 2 , wherein said inherent inaccuracy is associated with an inherent inaccuracy in determining the exact geo-spatial position of the lidar sensor.
5 . The system of claim 2 , wherein said inherent inaccuracy is associated with movement of the on-road vehicle while creating said representation.
6 . The system of claim 2 , wherein said inherent inaccuracy is associated with a failure to capture sufficient information related to the object in conjunction with image capturing performed by only one of the on-road vehicles.
7 . The system of claim 1 , wherein the plurality of data gathering configurations are a plurality of multi-camera vision configurations.
8 . The system of claim 1 , wherein the representations comprises representation of static objects in said areas.
9 . The system of claim 8 , wherein the static objects are associated with at least one of: (i) buildings, (ii) traffic signs, (iii) roads, (iv) trees, and (v) road hazards such as pits.
10 . The system of claim 1 , wherein said data combining technique is associated with at least one of: (i) statistical averaging, (ii) inverse convolution, (iii) least-mean-squares (LMS) algorithms, (iv) monte-carlo methods, and (v) application of machine learning models.
11 . The system of claim 1 , wherein said inherent inaccuracy is manifested as a reduced resolution of the representation.
12 . The system of claim 1 , wherein said inherent inaccuracy is manifested as a reduced accuracy in geo-spatial positioning of elements in the representation.
13 . The system of claim 1 , wherein said inherent inaccuracy is manifested as a reduced machine-learning model complexity associated with the representation.
14 . A method for utilizing a plurality of on-road vehicles for increasing descriptive accuracy of areas and/or objects, comprising:
in a post-capture analysis, receiving, in a server, a plurality of descriptions respectively from a plurality of on-road vehicles moving in visual vicinity of said particular area and/or an object, in which each of the descriptions having an inherent descriptive inaccuracy;
correlating, by the server, at least two of the received descriptions of a same area and/or object, wherein said at least two descriptions were captured by different ones of said plurality of on-road vehicles at different times; and
processing, in the server, said correlated descriptions, into a fused description, using at least one data combining techniques, in which said fused description has a better descriptive accuracy.