US-12619250-B2 - Ultrasonic piezoelectric transceiver sensor for full surface contact localization

US12619250B2US 12619250 B2US12619250 B2US 12619250B2US-12619250-B2

Abstract

The present disclosure provides methods, apparatuses, systems, and computer-readable mediums for classifying a region and an intensity of a contact. A method includes obtaining, from a plurality of acoustic sensors provided on an inner surface of a bumper of the apparatus, a combined acoustic signal, the combined acoustic signal being based on an input signal provided to the plurality of acoustic sensors, determining, using a trained machine learning model that has been trained with acoustic signals and corresponding position and intensity information of impacts on a plurality of regions of an outer surface of the bumper, the region of the contact on the bumper with respect to the plurality of regions and the intensity of the contact, based on the combined acoustic signal, and determining a motion of the apparatus based on the region of the contact, the intensity of the contact, and an operating mode of the apparatus.

Inventors

Adarsh K. KOSTA
Alexis M. Burns
Caleb Escobedo
Siddharth RUPAVATHARAM
Richard E. Howard
Lawrence Jackel
Daewon Lee
Ibrahim Volkan Isler

Assignees

SAMSUNG ELECTRONICS CO., LTD.

Dates

Publication Date: 20260505
Application Date: 20231215

Claims (20)

1 . A method for classifying a region and an intensity of a contact by an apparatus, comprising: determining an excitation frequency of an input signal that corresponds to an operating mode of the apparatus; generating mechanical vibrations on a portion of the apparatus by providing the input signal to each of a plurality of acoustic sensors, the plurality of acoustic sensors provided on an inner surface of a bumper of the apparatus, the plurality of acoustic sensors attached apart from each other; obtaining, from the plurality of acoustic sensors, a combined acoustic signal, the combined acoustic signal being based on the input signal; extracting signal magnitudes corresponding to the combined acoustic signal based on comparing the combined acoustic signal with a reference acoustic signal from a reference acoustic sensor, wherein the reference acoustic signal is generated by providing the input signal to the reference acoustic sensor; determining, using a trained machine learning model that has been trained with acoustic signals and corresponding position and intensity information of impacts on a plurality of regions of an outer surface of the bumper, the region of the contact on the bumper with respect to the plurality of regions and the intensity of the contact, based on the combined acoustic signal; and determining a motion of the apparatus based on the region of the contact, the intensity of the contact, and the operating mode of the apparatus.
2 . The method of claim 1 , wherein the plurality of acoustic sensors comprise a plurality of piezoelectric transceivers, and wherein the obtaining of the combined acoustic signal is performed by the plurality of piezoelectric transceivers.
3 . The method of claim 1 , wherein the plurality of acoustic sensors further comprise a bridge circuit, wherein the extracting of the signal magnitudes comprises inputting the combined acoustic signal and the reference acoustic signal to the bridge circuit to obtain a difference signal between the combined acoustic signal and the reference acoustic signal, and wherein the difference signal comprises the signal magnitudes corresponding to the combined acoustic signal.
4 . The method of claim 1 , wherein the operating mode of the apparatus is a contact avoidance mode, and wherein the determining of the motion of the apparatus comprises updating a path plan of the apparatus to move away from the contact, based on the region of the contact and the intensity of the contact.
5 . The method of claim 1 , wherein the operating mode of the apparatus is an edge following mode, and wherein the determining of the motion of the apparatus comprises: determining, based on the intensity of the contact, whether at least one of a light contact and a heavy contact is detected; based on the heavy contact being detected, adjusting a path plan of the apparatus to move towards a previous region where a previous light contact was previously detected; based on the light contact being detected, determining whether the region of the contact corresponds to the path plan of the apparatus; and based on the region of the contact not corresponding to the path plan of the apparatus, adjusting the path plan of the apparatus to move towards the previous region where the previous light contact was previously detected.
6 . The method of claim 1 , wherein the obtaining of the combined acoustic signal comprises: providing the input signal to each of the plurality of acoustic sensors, wherein the combined acoustic signal is received by the plurality of acoustic sensors while each of the plurality of acoustic sensors transmit the input signal provided to each of the plurality of acoustic sensors.
7 . The method of claim 6 , further comprising: providing the plurality of acoustic sensors with an input calibration signal corresponding to the operating mode of the apparatus; obtaining at least one calibration acoustic signal from the plurality of acoustic sensors; extracting calibration signal magnitudes from the at least one calibration acoustic signal; selecting, based on the calibration signal magnitudes, a calibration frequency having a maximum spectral power; and updating the excitation frequency of the input signal corresponding to the operating mode of the apparatus, based on the calibration frequency.
8 . The method of claim 6 , further comprising: changing the operating mode of the apparatus from a first mode to a second mode; and changing the excitation frequency of the input signal from a first excitation frequency to a second excitation frequency that corresponds to the second mode.
9 . The method of claim 6 , wherein the obtaining of the combined acoustic signal further comprises: providing the input signal to the reference acoustic sensor different from the plurality of acoustic sensors; and comparing the combined acoustic signal with the reference acoustic signal from the reference acoustic sensor, resulting in at least one difference signal.
10 . The method of claim 9 , further comprising: converting the at least one difference signal to at least one digital difference signal; phase matching the at least one digital difference signal; mixing the at least one phase-matched digital difference signal; and providing the at least one mixed phase-matched digital difference signal to the trained machine learning model as input.
11 . The method of claim 1 , wherein the trained machine learning model has been trained using contact sample data comprising a plurality of impact intensities at regions corresponding to the plurality of regions of the outer surface of the bumper on a plurality of sensing materials, and wherein the determining of the region of the contact on the bumper and the intensity of the contact comprises obtaining, from the trained machine learning model, a mapping of the region of the contact to at least one region of the plurality of regions and a classification of the intensity of the contact.
12 . An apparatus for classifying a region and an intensity of a contact, comprising: a plurality of piezoelectric transceivers provided on an inner surface of a bumper of the apparatus, the plurality of piezoelectric transceivers being attached apart from each other; a memory storing instructions; and one or more processors communicatively coupled to the memory; wherein the one or more processors are configured to execute the instructions to: determining an excitation frequency of an input signal that corresponds to an operating mode of the apparatus; generate mechanical vibrations on a portion of the apparatus by providing the input signal to each of the plurality of piezoelectric transceivers; obtain, from each of the plurality of piezoelectric transceivers, at least one acoustic signal, the at least one acoustic signal being based on an input the input signal provided to each of the plurality of piezoelectric transceivers; extract signal magnitudes corresponding to the at least one acoustic signal based on comparing each acoustic signal of the at least one acoustic signal from a corresponding piezoelectric transceiver of the plurality of piezoelectric transceivers with a reference acoustic signal, wherein the reference acoustic signal is generated by providing the input signal to a reference piezoelectric transceiver; determine, using a trained machine learning model that has been trained with acoustic signals and corresponding position and intensity information of impacts on a plurality of regions of an outer surface of the bumper, the region of the contact on the bumper with respect to the plurality of regions and the intensity of the contact, based on the signal magnitudes; and determine a motion of the apparatus based on the region of the contact, the intensity of the contact, and the operating mode of the apparatus.
13 . The apparatus of claim 12 , wherein the apparatus further comprises a bridge circuit, wherein the one or more processors are further configured to execute further instructions to input the at least one acoustic signal and the reference acoustic signal to the bridge circuit to obtain a difference signal between each of the at least one acoustic signal and the reference acoustic signal, and wherein the difference signal comprises the signal magnitudes corresponding to the at least one acoustic signal.
14 . The apparatus of claim 12 , wherein the operating mode of the apparatus is a contact avoidance mode, and wherein the one or more processors are further configured to execute further instructions to update a path plan of the apparatus to move away from the contact, based on the region of the contact and the intensity of the contact.
15 . The apparatus of claim 12 , wherein the operating mode of the apparatus is an edge following mode, and wherein the one or more processors are further configured to execute further instructions to: determine, based on the intensity of the contact, whether at least one of a light contact and a heavy contact is detected; based on the heavy contact being detected, adjust a path plan of the apparatus to move towards a previous region where a previous light contact was previously detected; based on the light contact being detected, determine whether the region of the contact corresponds to the path plan of the apparatus; and based on the region of the contact not corresponding to the path plan of the apparatus, adjust the path plan of the apparatus to move towards the previous region where the previous light contact was previously detected.
16 . The apparatus of claim 12 , wherein the one or more processors are further configured to execute further instructions to: provide the input signal to each of the plurality of piezoelectric transceivers, wherein the at least one acoustic signal is received by the plurality of piezoelectric transceivers while each of the plurality of piezoelectric transceivers transmit the input signal provided to each of the plurality of piezoelectric transceivers.
17 . The apparatus of claim 16 , wherein the one or more processors are further configured to execute further instructions to: provide the plurality of piezoelectric transceivers with an input calibration signal corresponding to the operating mode of the apparatus; obtain at least one calibration acoustic signal from the plurality of piezoelectric transceivers; extract calibration signal magnitudes from the at least one calibration acoustic signal; select, based on the calibration signal magnitudes, a calibration frequency having a maximum spectral power; and update the excitation frequency of the input signal corresponding to the operating mode of the apparatus, based on the calibration frequency.
18 . The apparatus of claim 16 , wherein the one or more processors are further configured to execute further instructions to: provide the input signal to the reference piezoelectric transceiver different from the plurality of piezoelectric transceivers; compare each of the at least one acoustic signal with the reference acoustic signal from the reference piezoelectric transceiver, resulting in at least one difference signal; convert the at least one difference signal to at least one digital difference signal; phase match the at least one digital difference signal; mix the at least one phase-matched digital difference signal; and provide the at least one mixed phase-matched digital difference signal to the trained machine learning model as input.
19 . The apparatus of claim 12 , wherein the plurality of piezoelectric transceivers comprises a number of piezoelectric transceivers determined based on at least one of an accuracy threshold of the apparatus and a material type of the bumper of the apparatus, and wherein the plurality of piezoelectric transceivers are provided at locations on the inner surface of the bumper of the apparatus determined based on the at least one of the accuracy threshold of the apparatus and the material type of the bumper of the apparatus.
20 . A method for classifying a region and an intensity of a contact by an apparatus, comprising: determining an excitation frequency of an excitation signal that corresponds to an operating mode of the apparatus; generating mechanical vibrations on a portion of the apparatus by providing the excitation signal to each of a plurality of acoustic transceivers and to a reference acoustic transceiver, the plurality of acoustic transceivers being provided on an inner surface of a bumper of the apparatus, the plurality of acoustic transceivers attached apart from each other; obtaining, from each of the plurality of acoustic transceivers, at least one acoustic signal, the at least one acoustic signal being received by the plurality of acoustic transceivers while each of the plurality of acoustic transceivers transmit the excitation signal provided to each of the plurality of acoustic transceivers; extracting signal magnitudes corresponding to the at least one acoustic signal based on comparing each of the at least one acoustic signal from a corresponding acoustic transceiver of the plurality of acoustic transceivers based on a reference signal from the reference acoustic transceiver, wherein the reference signal is generated by the reference acoustic transceiver using the excitation signal; and estimating, using a trained machine learning model based on the signal magnitudes, the region of the contact from among a plurality of contact regions on an outer surface of the bumper and the intensity of the contact from among a plurality of force levels, the trained machine learning model having been trained with acoustic signals and corresponding position and intensity information of impacts on the plurality of contact regions on the outer surface of the bumper.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application claims benefit of priority under 35 U.S.C. § 119 to U.S. Provisional Patent Application No. 63/449,910, filed on Mar. 3, 2023, and U.S. Provisional Patent Application No. 63/531,642, filed on Aug. 9, 2023, in the U.S. Patent and Trademark Office, the disclosure of which is incorporated by reference herein in its entirety. BACKGROUND 1. Field The present disclosure relates generally to robotic control, and more particularly to methods, apparatuses, systems, and non-transitory computer-readable mediums for classifying a region and an intensity of a contact with a full-surface contact localization system using ultrasonic piezoelectric transceiver sensors. 2. Description of Related Art As robots enter dynamic spaces where obstacles such as, but not limited to, people, animals, and/or objects, may be present, making physical contact with the obstacles may be inevitable. Consequently, robots may need tactile sensing capabilities that may detect when and where a contact occurs in order to navigate these dynamic spaces. That is, providing robots with full-surface contact detection and localization sensing capabilities may provide the robots with a capability of perceiving and/or avoiding obstacles post-contact, which may also contribute to information related to the robot's surroundings (e.g., area and/or volume within a threshold distance). Related approaches for emulating human skin and/or touch receptors may be based on capacitive, optical, resistive, and/or vision approaches. However, each of these approaches may have disadvantages that may prevent wider adoption of these approaches. For example, capacitance-based approaches are well suited for contact detection and localization of point contacts when multiple capacitive pressure sensors are deployed as a matrix that covers the surface. Alternatively or additionally, the electrical properties of an object (e.g., response to an electrical field) may affect the detection of a contact with a capacitance-based sensor. For example, electrically-conducting objects (e.g., metals) may be easier to detect than objects made of other materials (e.g., non-conductive materials). Furthermore, localization resolution may depend on sensor size, and as such, a greater number of sensors may be needed in order to realize a desired localization resolution. As another example, optical and/or camera-based approaches may have a fixed field of view that may limit sensing areas, and as such, may necessitate a dense sensor deployment. Alternatively or additionally, performance of camera-based approaches may degrade when faced with visual occlusions, poor light conditions, and/or transparent or mirrored objects that may be difficult to detect visually. Furthermore, camera-based approaches may typically not be accurate over very short ranges (e.g., less than 10 centimeters (cm)) depending on camera focal length. As another example, proprioceptive (resistive) approaches may use sensors (e.g., joint torque sensors, motor torque sensors, position sensors, velocity sensors, momentum sensors, and the like) coupled with inverse kinematics and/or dynamics to infer a contact region and/or a contact force. However, these proprioceptive approaches may provide poor estimation results of contact region and/or contact force due to noisy and/or time varying properties of the motors. Thus, there exists a need for further improvements to tactile sensing approaches, as the need for navigating in dynamic spaces may be constrained by an inability to assess a region and an intensity of a contact. Improvements are presented herein. These improvements may also be applicable to other movement control and/or mapping technologies. SUMMARY The following presents a simplified summary of one or more embodiments of the present disclosure in order to provide a basic understanding of such embodiments. This summary is not an extensive overview of all contemplated embodiments, and is intended to neither identify key or critical elements of all embodiments nor delineate the scope of any or all embodiments. Its sole purpose is to present some concepts of one or more embodiments of the present disclosure in a simplified form as a prelude to the more detailed description that is presented later. Methods, apparatuses, systems, and non-transitory computer-readable mediums for classifying a region and an intensity of a contact are disclosed by the present disclosure. Aspects of the present disclosure provide for determining a motion based on the region of the contact, the intensity of the contact, and an operating mode. According to an aspect of the present disclosure, a method for classifying a region and an intensity of a contact by an apparatus is provided. The method includes obtaining, from a plurality of acoustic sensors provided on an inner surface of a bumper of the apparatus, a combined acoustic signal, the combined acoustic signal being based on an input signal