Material identification using vibration signals
View Patent ↗Described herein are systems, methods, and other techniques for determining a material type while an implement of a construction machine is interacting with a ground surface. A vibration signal that is indicative of a movement of the implement is captured. One or more features are extracted from the vibration signal. The one or more features are provided to a machine-learning model to generate a model output. The material type of the ground surface is predicted based on the model output.
1 . A computer-implemented method of determining a material type while an implement of a construction machine is interacting with a ground surface, the computer-implemented method comprising:
causing a movement of the implement using one or more control signals;
capturing a vibration signal that is indicative of the movement of the implement;
extracting one or more features from the vibration signal;
estimating one or more positions of the implement during the movement of the implement;
providing the one or more features and the one or more positions of the implement to a machine-learning model to generate a model output; and
predicting the material type of the ground surface based on the model output.
2 . The computer-implemented method of claim 1 , wherein the vibration signal is captured using a vibration sensor mounted to the construction machine.
3 . The computer-implemented method of claim 2 , wherein the vibration sensor includes one or both of an accelerometer or a gyroscope, and wherein the vibration signal includes one or both of an acceleration signal or a rotation signal.
4 . The computer-implemented method of claim 2 , wherein the vibration sensor is mounted to the implement.
5 . The computer-implemented method of claim 1 , wherein the one or more features include one or both of signal amplitude features or signal frequency features.
6 . The computer-implemented method of claim 1 , wherein the machine-learning model is a pre-trained artificial recurrent neural network, a feed-forward neural network, or a support-vector machine.
7 . The computer-implemented method of claim 1 , further comprising:
predicting a first material type of a first portion of the ground surface based on the model output; and
predicting a second material type of a second portion of the ground surface based on the model output.
8 . The computer-implemented method of claim 1 , further comprising:
predicting a location of a boundary between a first material type of a first portion of the ground surface and a second material type of a second portion of the ground surface based on the model output.
9 . A system for determining a material type while an implement of a construction machine is interacting with a ground surface, the system comprising:
one or more processors; and
a computer-readable medium comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:
causing a movement of the implement using one or more control signals;
capturing a vibration signal that is indicative of the movement of the implement;
extracting one or more features from the vibration signal;
estimating one or more positions of the implement during the movement of the implement;
providing the one or more features and the one or more positions of the implement to a machine-learning model to generate a model output; and
predicting the material type of the ground surface based on the model output.
10 . The system of claim 9 , wherein the vibration signal is captured using a vibration sensor mounted to the construction machine.
11 . The system of claim 10 , wherein the vibration sensor includes one or both of an accelerometer or a gyroscope, and wherein the vibration signal includes one or both of an acceleration signal or a rotation signal.
12 . The system of claim 10 , wherein the vibration sensor is mounted to the implement.
13 . The system of claim 9 , wherein the one or more features include one or both of signal amplitude features or signal frequency features.
14 . The system of claim 9 , wherein the machine-learning model is a pre-trained artificial recurrent neural network, a feed-forward neural network, or a support-vector machine.
15 . The system of claim 9 , wherein the operations further comprise:
predicting a first material type of a first portion of the ground surface based on the model output; and
predicting a second material type of a second portion of the ground surface based on the model output.
16 . The system of claim 9 , wherein the operations further comprise:
predicting a location of a boundary between a first material type of a first portion of the ground surface and a second material type of a second portion of the ground surface based on the model output.
17 . A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors, cause the one or more processors to perform operations for determining a material type while an implement of a construction machine is interacting with a ground surface, the operations comprising:
causing a movement of the implement using one or more control signals;
capturing a vibration signal that is indicative of the movement of the implement;
extracting one or more features from the vibration signal;
estimating one or more positions of the implement during the movement of the implement;
providing the one or more features and the one or more positions of the implement to a machine-learning model to generate a model output; and
predicting the material type of the ground surface based on the model output.
18 . The non-transitory computer-readable medium of claim 17 , wherein the vibration signal is captured using a vibration sensor mounted to the construction machine.
19 . The non-transitory computer-readable medium of claim 18 , wherein the vibration sensor includes one or both of an accelerometer or a gyroscope, and wherein the vibration signal includes one or both of an acceleration signal or a rotation signal.
20 . The non-transitory computer-readable medium of claim 18 , wherein the vibration sensor is mounted to the implement.