WO2025213279A1

WO2025213279A1 - Robotic structural sensing and protective systems and methods

Info

Publication number: WO2025213279A1
Application number: PCT/CA2025/050537
Authority: WO
Inventors: Nima ZAMANI; Dan Vu; Kareem ALASWAD
Original assignee: Cobionix Corp
Current assignee: Cobionix Corp
Priority date: 2024-04-11
Filing date: 2025-04-11
Publication date: 2025-10-16
Anticipated expiration: 2026-10-11
Also published as: WO2025213278A1

Abstract

Provided is a shape memory alloy (SMA) protective component for protecting structural or functional components in a robot apparatus and/or support structures the robot apparatus is mounted to. The protective component is generally disposed within a limb segment of a robot apparatus, between two joints/servos. According to some embodiments, the protective component is itself a limb segment, and two or more protective components can be connected at joints, like building blocks, to form the robot apparatus. The protective component includes one or more "smart" shape memory alloy structures that are deformable to shield/absorb external impacts and dissipate mechanical energy. The material of the shape memory alloy is conductive, flexible, and malleable such that it can be interwoven into a structural shape or pattern. After deforming from an impact, passing electricity through the shape memory alloy structure causes the structure to inflate/reform its original shape.

Description

ROBOTIC STRUCTURAL SENSING AND PROTECTIVE SYSTEMS AND METHODS

Technical Field

[0001] The embodiments disclosed herein relate to robotics, and, in particular to protective component structural components employing shape memory alloys for impact absorption, self-reforming capabilities, and integrated sensing in robotic systems.

Introduction

[0002] Robotics for human interaction must be carefully configured to prevent injury or discomfort to humans during interactions. This is particularly true for healthcare applications where a robot comes into direct physical contact with a patient who may have limited mobility, is injured, is sedated, etc. Balancing a robot’s freedom of movement to perform operations with patient safety is therefore an important consideration for robotics in a healthcare setting.

[0003] Over time, robotic systems have evolved from simple, rigid frameworks to more adaptive designs capable of interacting safely with humans and their surrounding environment. However, many traditional robotic limbs are still built around metals and composites that, while providing structural rigidity, lack the ability to dynamically absorb and dissipate impact forces. This rigidity creates operational vulnerabilities, including potential damage when abrupt impacts occur, as well as increased downtime and maintenance costs.

[0004] To address these limitations, recent innovations have introduced shape memory alloy structures integrated into robotic limbs. These “smart” materials are capable of deforming to absorb external forces — similar to automotive crumple zones — and then returning to their original form once the impact subsides or a current is applied. In addition, embedding sensors within these alloys makes it possible to measure deformation, force distribution, and other critical data in real time, thus enabling sophisticated feedback loops for enhanced safety and efficient energy use.

[0005] By pairing shape memory alloys with conductive sleeves and flexible PCB components, robots can autonomously restore their structural integrity without requiring manual repairs or part replacements. This transformative capability significantly reduces downtime, prolongs overall component life, and enables new use cases — from medical devices requiring precise haptic feedback to advanced industrial robots that must operate reliably in high-impact or unpredictable conditions.

[0006] Accordingly, there is a need for new protective components for robotics that have shape memory alloy functionality to permit close-in robot-human interactions.

Summary

[0007] Shape memory alloy (SMA) protective components, systems and methods are described. According to an embodiment, there is a shape memory protective component. The component includes opposed first and second servo motors, each motor having a hollow shaft. A tube connects the hollow shafts for routing cabling between the first and second servo motors via the hollow shafts. A SMA structure surrounds the tube. The SMA structure is configured to deform and dissipate energy by hysteresis in response to external force applied to the structure, and reform to an original shape when electric current is applied to the structure.

[0008] According to another embodiment, there is a robotic system comprising at least one limb segment comprising a SMA structure configured to deform and dissipate energy by hysteresis in response to external force applied to the SMA structure, and reform to an original shape when electric current is applied to the SMA structure. The robotic system further includes sensors configured for sensing an environment around the limb segment and a control system configured to receive input from the plurality of sensors to detect a potential collision between the robotic limb segment and an object in the environment and modulate current applied to the SMA structure in response to detecting the potential collision.

[0009] According to another embodiment, there is a method for vibration and impact mitigation. The method comprises providing an apparatus composing a SMA structure configured to deform and dissipate energy by hysteresis in response to external force applied to the SMA structure, and reform to an original shape when electric current is applied to the SMA structure. The method further includes constantly applying the electric current to the SMA structure to substantially maintain the SMA structure in the original shape and performing an activity which causes the apparatus to experience vibrations and impacts.

[0010] Other aspects and features will become apparent, to those ordinarily skilled in the art, upon review of the following description of some exemplary embodiments.

Brief Description of the Drawings

[0011] The drawings included herewith are for illustrating various examples of articles, methods, and apparatuses of the present specification and are not drawn to scale. In the drawings:

[0012] FIG. 1 is a perspective view of a robot apparatus, according to an embodiment;

[0013] FIG. 2A is cutaway view of a shape memory alloy protective component, according to an embodiment;

[0014] FIG. 2B is the protective component of FIG. 2A shown during a deformationreformation cycle;

[0015] FIGS. 20 and 2D cutaway views protective components, according to various embodiments; and

[0016] FIG. 3 is a diagram of control system, according to an embodiment.

Detailed Description

[0017] Various apparatuses or processes will be described below to provide an example of each claimed embodiment. No embodiment described below limits any claimed embodiment and any claimed embodiment may cover processes or apparatuses that differ from those described below. The claimed embodiments are not limited to apparatuses or processes having all of the features of any one apparatus or process described below or to features common to multiple or all of the apparatuses described below.

[0018] One or more systems described herein may be implemented in computer programs executing on programmable computers, each comprising at least one processor, a data storage system (including volatile and non-volatile memory and/or storage elements), at least one input device, and at least one output device. For example, and without limitation, the programmable computer may be a programmable logic unit, a mainframe computer, server, and personal computer, cloud based program or system, laptop, personal data assistance, cellular telephone, smartphone, or tablet device.

[0019] Each program is preferably implemented in a high level procedural or object oriented programming and/or scripting language to communicate with a computer system. However, the programs can be implemented in assembly or machine language, if desired. In any case, the language may be a compiled or interpreted language. Each such computer program is preferably stored on a storage media or a device readable by a general or special purpose programmable computer for configuring and operating the computer when the storage media or device is read by the computer to perform the procedures described herein.

[0020] A description of an embodiment with several components in communication with each other does not imply that all such components are required. On the contrary a variety of optional components are described to illustrate the wide variety of possible embodiments of the present invention.

[0021] Further, although process steps, method steps, algorithms or the like may be described (in the disclosure and I or in the claims) in a sequential order, such processes, methods and algorithms may be configured to work in alternate orders. In other words, any sequence or order of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order that is practical. Further, some steps may be performed simultaneously.

[0022] When a single device or article is described herein, it will be readily apparent that more than one device I article (whether or not they cooperate) may be used in place of a single device I article. Similarly, where more than one device or article is described herein (whether or not they cooperate), it will be readily apparent that a single device I article may be used in place of the more than one device or article. [0023] The terms “controller” and “compute unit” are used interchangeably herein to refer to computer processors (e.g., central processing units, graphics processing units), integrated circuits, systems-on-a-chip, including associated hardware and software, being configured to execute instructions, perform calculations and/or process signals/data as the case may be, unless otherwise specified.

[0024] A shape memory alloy is a material which can be deformed at a certain temperature or temperature range, and return to its original shape when heated. A desired original (undeformed) shape of the material can be “pre-programmed” by subjecting the material to thermal treatment and geometric shaping using methods known in the art. Herein, “shape memory alloy” and “SMA” refer to electrically conductive material which is heated by applying an electric current to the material.

[0025] Herein, the terms “inflate,” “reform” and conjugations thereof, refer to SMA material returning to its original shape after deformation. Herein, the term “stiffen” and conjugations thereof refer to a state where an electric current is applied to a SMA to reform its original shape after deformation, or to resist deformation by substantially maintaining its original shape.

[0026] Herein, the term “deflate” and conjugations thereof, refer to SMA material deforming in response to applied force or impact. Herein, the term “soften” and conjugations thereof refer to a state where an electric current is removed or not applied to a SMA to allow the SMA to deform in response to applied forces.

[0027] Referring to FIG. 1 , shown therein is a perspective view of an autonomous robot apparatus 100, according to an embodiment. The robot apparatus 100 may be the robot apparatus described in United States Patent Publication No. 20220362944, to the same applicant, the entirety of which is incorporated herein by reference.

[0028] The robot 100 includes an articulated arm 102. The arm 102 includes limb segments 104a, 104b, 104c, 104d connected at joints 106a, 106b, 106c. The limb segments 104b, 104c, 104d are rotatable about the joints 106a, 106b, 106c. The limb segment 104a is rotatable about a base 120. The connectivity of the limb segments 104a, 104b, 104c, 104d via the joints 106a, 106b, 106c provides for the robot 100 to articulate the arm 102 in three dimensions. [0029] Each limb segment 104a, 104b, 104c, 104d is preferably fabricated using primarily carbon fiber composite material to be relatively light weight and resilient. According to other embodiments the limb segments 104a, 104b, 104c, 104d may be constructed of other composite materials such as fiberglass.

[0030] Each limb segment 104a, 104b, 104c, 104d houses one or more servo motors for actuating the limb segment. All wiring to the servo motors is routed through the interior of the limb segments 104a, 104b, 104c, 104d and through the hollow joints 106a, 106b, 106c, with no wiring or cables present on the exterior of the robot 100 which can limit installation and movement of the robot 100 during operation.

[0031] The robot 100 includes a base 120. The base 120 is fixed to a support structure (not shown) for supporting the arm 102. Generally, the base 120 includes a central compute unit for overall control of the robot 100. According to some embodiments, the base 120 is mounted to a cart or other support structure.

[0032] According to other embodiments, the robot 100 includes two substantially similar arms, both connected to the base 120. Each arm may be configured for specific tasks, as described below.

[0033] The robot 100 includes computational power as well as sensors for autonomous applications. The sensors include external sensors including vision systems such as LIDAR, time of flight (TOF) and stereoscopic depth sensors. Some of the sensors are stationary at the base 120 of the robot 100 to sense and generate information regarding the environment the robot 100 is operating in. According to an embodiment, the base 120 includes a rotating lidar system or multiple stationary depth sensors located at specific separation angles.

[0034] Other sensors are mounted to the arm 102 of the robot 100, for example, the proximity sensing skin disclosed in United States Patent Publication No. 20220362944.

[0035] Sensors are also included in vision modules 108a, 108b in the base and on the end of the robot arm 102. The vision module 108a in the base includes 2 RGB cameras and other sensors on a 360-degree gimbal system with pitch, yaw and roll control allowing for the sensors to detect and focus in on different objects. The vision module 108b on the end of the robot arm includes 2 stereoscopic cameras, a time of flight sensor, a thermal pile sensor, an inertial measurement unit and a projector. The vision module 108 and the sensors therein may be the vision module disclosed in United States Provisional Patent Application No. 63/507898, to the same applicant, the entirety of which is incorporated herein by reference.

[0036] The sensors are wholly incorporated into the robot 100 and are meant to be used out of the box as a cohesive unit in conjunction with dedicated software. The robot 100 also comes installed with Al for various autonomous tasks fully integrated and ready to use. A human operator or patient can physically touch the body of the robot and with its capacitive and time of flight sensors combined with the vision system data, and proximity sensing skin can recognize and detect the incoming touch. Upon physical contact, the robot can enter a compliant mode where the users can manipulate the linkages/joints/body of the robot.

[0037] The robot 100 includes a mounting interface 110 for removably attaching tools (not shown). Additional sensors can be incorporated into the tool connected to the mounting interface 110 for application-specific needs.

[0038] The robot 100 includes a fluid handling system comprising pneumatic and/or hydraulic pumps or compressors in a base or support structure and fluid lines passing through the arm to the mounting interface 110. The fluid handling system is configured for dispensing gas or liquid through an end effector or tool attached to the mounting interface 110 and connected to the fluid lines.

[0039] Referring to FIG. 2A, shown therein is a shape memory alloy (SMA) protective component 200, according to an embodiment. Generally, the protective component 200 is provided to shield robotic components from impacts, collisions or external forces. In the embodiment shown, the protective component 200 is disposed between two joints/servos 202 of a robot apparatus. The protective component 200 may be wholly contained within or surrounding a limb segment of a robot apparatus (e.g., limb segment 104c in robotic arm 102), between two joints/servos. According to some embodiments, the protective component 200 itself is a limb segment for a robot apparatus, and two or more protective components 200 can be connected at joints, like building blocks, to form the robot apparatus. As shown, the protective component 200 is bent/curved. According to other embodiments, the protective component 200 may be straight, with both servos 202a, 202b in-line.

[0040] The protective component 200 includes one or more “smart” SMA structures 220 that are deformable and inflatable materials used to shield/absorb external impacts and dissipate mechanical energy waste by hysteresis. This process is akin to a vehicle’s crumple zone that deforms to absorb impact forces to reduce the forces felt by the vehicle occupants. In the present context, the protective component 200 protects other structural or functional components in the interior of the robot apparatus and/or support structures the robot apparatus is mounted to.

[0041] Referring to FIG. 2B, in a neutral state, with no external forces applied, the SMA structure 220 can readily absorb impact and deform. After impact, when the SMA structure 220 is deformed, the processes described below enable the smart material to be inflated or reformed into the original neutral state/shape, ready to absorb impact again and continuously repeat this process until the material’s useful lifecycle is depleted.

[0042] Referring again to FIG. 2A, the protective component 200 includes 2 servo motors 202 with at least a first servo motor 202a having a hollow shaft 204a. In a preferred embodiment, both servo motors 202a, 202b, include hollow shafts 204a, 204b. In some embodiments, each servo motor 202 is disposed at a joint in a robotic arm (hereinafter the servo motors and joints are collectively referred to as servos/joints 202).

[0043] A rigid tube 206 connects the hollow shaft 204a of the first motor 202a to the second servo motor 202b creating a bridge connecting the servos/joints 202 together. Preferably, the tube 206 is constructed of and includes graphene particles filled with epoxy for increased stiffness. Various cables, not limited to power, electrical, or fluid lines are routed through the tube 206 and the hollow shafts 204.

[0044] Each servo motor 202 includes a printed circuit board (PCB) 208 that is mounted in place on the servos/joints 202 creating an additional layer. The PCB 208 may be a flexible PCB. The layer or PCB 208 has several nodes/connectors/inputs for electrical and/or data connections. A housing (not shown) covers the PCB 208 and attaches to the servo motors 202. The housing is designed to leave the electrical contacts on the PCB 208 exposed for connecting the nodes/connectors/inputs to the PCB 208. According to various embodiments, the housing may be constructed of aluminum, steel, carbon fiber or 3D-printed materials.

[0045] In addition to the PCB 208 being physically mounted to the surface of the servo/joint 202, there is an electrical and a data connection to the joints via cabling, pins, and other modalities to create an electronic data/power exchange between the servo/joint 202 and the PCB 208. As such, the servo/joints 202 are connected to a power source through cables or other modalities of electronic-data-power transmission. The cables or other methods can transmit exclusively power or exclusively data or a combination thereof.

[0046] The protective component 200 includes a volumetric area 210 bounded by a sleeve 212 surrounding the carbon fibre tube 206 connecting the two servos/joints 202. A shape memory alloy structure 220 is disposed within the volumetric area 210. The material of the shape memory alloy 220 is conductive, flexible, and malleable such that it can be interwoven into a structural shape or pattern. According to some embodiments, the shape memory alloy structure 220 is provided in the form of a large sheet but is not limited to such a shape or pattern.

[0047] In the embodiments shown in FIGS. 2A-2C, the shape memory alloy structure 220 is configured as a mesh or lattice structure constructed of individual SMA elements 222. The SMA element may be, a single SMA strand having two endpoints or nodes. Each endpoint or node is connected to a connector or access point on at least one of the PCBs 208 mounted on the servos/joints 202. This allows the shape memory alloy 220 to access a source of electricity to activate its physical properties for inflating/deflating or deform ing/reforming. The strands 222 forming the shape memory alloy structure 220 draw electricity from the PCBs 208 to activate in unison, or individually, thereby causing the shape memory alloy structure 220 to inflate/deflate or deform/reform. According to other embodiments, the SMA element 222 may be a cluster of localized SMA strands, such as wires and strips that are interwoven or arranged into a 3D structure. [0048] The strands 222 are not limited to the direction or orientation in which they are connected to the PCBs 208. The strands 222 can be interwoven to create different structural patterns that are not limited to: Garter stich; Moss Stich; Rib Stich; Cable Stich; Herringbone Stich; Waffle Stich; and other patterns overlapping vertically, horizontally and/or diagonally.

[0049] The strands 222 and shape memory alloy structure 220 may be constructed by 3D printing conductive materials for smart alloys. 3D printing provides for a variety of structures and shapes that the shape memory alloy structure 220 can take to substantially fill in the volumetric area 210 not occupied by the carbon fiber tube 206. The 3D printed materials will have endpoints/nodes/connectors that can be connected directly to the input nodes/slots on the PCBs 208 mounted on the servos/joints 202. Likewise, the overall shape memory alloy structure 220 that is printed can have slots/openings/nodes to allow for external electrical/power peripherals to be inserted to allow power to flow into the structure.

[0050] According to some embodiments, in addition to the overall SMA structure 220 itself being defeatable or deformable to absorb impact, the SMA structure 220 can be designed to incorporate various holes, openings and/or trenches that pass through between opposable ends of the SMA structure 220 to allow for secondary materials to pass through. According to various embodiments, the secondary materials can include the same shape memory alloy material the SMA structure 220 is constructed of, or other materials that are conductive, impact absorbent, and/or inflatable/deflatable with electrical current, to further assist the SMA structure 220 to absorb more impact or better return to its original shape. According to some embodiments the secondary material may be a polymer or rubberized plastic.

[0051] Now referring to FIG. 2C, shown therein is a protective component 250, according to another embodiment. The protective component 250 is substantially similar to the protective component 200 in FIG. 2A and further includes a plastic lattice structure 252. The plastic lattice 252 supports the SMA structure 220 and also provides additional impact/force absorption ability to the protective component 250. The plastic lattice 252 fills substantially the entire volumetric area 210 bounded by the sleeve 212 that is not occupied by the SMA structure 220 and the other components 202, 204, 206, 208 described above.

[0052] The protective component 260 in FIG. 2D is substantially similar to the protective component 250, but lacks the SMA structure 220. In this embodiment, the plastic lattice 254 provides all impact/force absorption for the protective component. Given the absence of SMA structure, the protective component 260 may not reform to its original shape to the same degree as the embodiments shown in FIGS. 2A-2C.

[0053] Referring to FIGS. 2C-2D, during 3D printing of the plastic lattice structure 252, 254, as strands or layers are starting to form, additional shape memory alloy strands can be placed into preplanned trenches in the partially printed structure where they can be laid to bond or fit within the final print. Any unoccupied space in the structure 220 not filled by the materials mentioned above will be air, padding, cables.

[0054] The sleeve 212 is a conductive fabric or mesh that can optionally cover the structure 220. According to some embodiments, the sleeve 212 is constructed using flexible PCB material, for example, STRETCH. flex by Wurth Elektronik. The sleeve 212 can partially or completely cover the shape memory alloy structure 220, the servos/joints 202 and the carbon fiber tube 206. According to various embodiments, the sleeve 212 can be a standalone component or combined with the shape memory alloy structure 220 to form a contiguous part. When combined with the structure 220, the endpoints I nodes of the sleeve 212 are connect to the PCBs 208 through the inputs on the PCBs 208 to transmit power or data omnidirectionally.

[0055] The physical properties of the malleable and conductive material in the shape memory alloy structure 220 and/or sleeve 212, provide for the following properties. [0056] In a neutral (rest) state where no external force is being applied, the shape memory alloy structure 220 can absorb external impact forces by its collapsible, deformable, and malleability physical properties.

[0057] By passing an electrical current between the two nodes of the strands 222 in the shape memory alloy structure 220, heat is generated to activate the conductive properties of the conductive material causing it to expand and reform its original rest state shape. [0058] During operation of high vibrational tasks, electrical current can be constantly applied to the shape memory alloy structure 220 to “prop up” the structure 220 to substantially maintain its original (undeformed) shape to continuously absorb external forces and apply outward force to the joint.

[0059] By calculating the amount of electrical resistance to the shape memory alloy structure 220, and the compute unit orchestrating and connecting the various components together, it can be determined how much the shape memory alloy structure 220 has been deformed, and where deformation has occurred, if at all. The degree of deformation can be correlated to the amount of external force being applied to cause the deformation.

[0060] External force measurement can be performed when the shape memory alloy structure 220 is at rest I in its neutral state and a current is constantly applied. When an impact occurs, there is a fluctuation to the constant current being applied. The current fluctuations in individual strands/nodes will have different values and through individual and agreement measurements of the nodes, the occurrence of an impact, the location/position of the impact, and how much energy was absorbed can be determined. Likewise, in a deflated/absorbed state, measurement of impact force and location can occur by applying and measuring how much current is being sent to each strand/node individually and in aggregate.

[0061] Additional sensors may be embedded into the sleeve 212. For example, sensors such as an EKG, an ECG, a thermocouple (or other temperature sensor) and Blood Oxygen can be embedded into an exterior-facing surface of the the sleeve 212 to receive external input from humans. For example, a user/patient may touch a finger to an embedded sensor in the sleeve 212 for vital sign measurements. In addition, strain, pressure and humidity sensors can be embedded into the sleeve 212 to sense the surrounding environment or forces applied to the sleeve 200.

[0062] Referring to FIG. 3, shown therein is a control system 400 of a robotic system, according to an embodiment. The robotic system may include the robot apparatus 100 shown in FIG. 1 and the SMA component 200 shown in FIG. 2. According to various embodiments, a central controller 402 of the robot apparatus implements one or more artificial intelligence (Al) modules 440, each module configured to interpret incoming sensor data, predict future states, and issue high-level directives to local controllers/microcontrollers 418 managing shape memory alloy actuation. These modules 440 can be deployed on standard central processing units (CPUs), graphics processing units (GPUs), tensor processing units (TPUs), or other specialized Al accelerators.

[0063] According to some embodiments, a multi-modal data fusion Al module 440 is provided. The Al module 440 receives streaming inputs from diverse sources: strain gauges and current/voltage sensors embedded between the SMA elements 422; the conductive sleeve 212 and the sensors embedded therein; traditional servo encoders in the servo motors 202; environmental sensors on the robot 100 (e.g., LIDAR, cameras, ultrasonic rangefinders); and operator commands. A deep neural network or similar algorithm may perform data fusion, filtering noise and correlating disparate signals to produce a unified representation of the robot’s internal and external states. This capability is beneficial for accurate detection of collisions, anomalies in mechanical loading on servos, joints or limbs of the robot 100, and subtle changes in the environment that might necessitate proactive reconfiguration.

[0064] According to some embodiments, to anticipate events that could compromise the robot’s structural integrity, the central compute unit 402 may employ reinforcement learning or supervised machine learning models trained on historical collision and deformation data. By analyzing patterns in sensor measurements (e.g., accelerometer readings, temperature spikes), or current draw profiles of SMA elements 422, the Al module 440 can identify precursors to high-impact events such as abrupt changes in the robot’s momentum or external objects moving into collision paths. Once a high-likelihood collision is predicted, the system 400 can initiate pre-emptive modulation of current applied to specific SMA elements 222, or the entire SMA structure 220 to “stiffen” or “soften” in order to optimize the robot’s stance to either absorb or mitigate the imminent force.

[0065] According to some embodiments, local microcontrollers 418 on the PCB 408 handle near-instantaneous shape adjustments by modulating current applied to SMA elements 222 in response to sensor feedback loops, while the Al module 440 of the central controller 402 supervises broader, system-level adaptation. For example, if the Al detects patterns indicating frequent collisions in a particular workspace region, it may adjust a global motion planning algorithm to reduce travel speeds or re-route the robot’s movement. This macro-level control can also integrate motion planning libraries that consider dynamic constraints such as joint torque limits, battery levels for SMA actuation, and thermal thresholds to avoid overheating the shape memory components 200, 220.

[0066] According to some embodiments, using anomaly detection techniques e.g., autoencoders, one-class Support Vector Machines, or statistical control charts, the Al identifies deviations in sensor data that deviate from established norms (e.g., persistent current spikes in a single SMA element 222). This proactive detection allows the robot to invoke preventive measures, such as redistributing loads among parallel/adjacent SMA elements, entering a low-power safe mode, or notifying maintenance personnel if a hardware fault is suspected. In severe cases, the Al can trigger an emergency reformation to protect critical components by dissipating or diverting forces away from vulnerable joints.

[0067] In some embodiments, the Al module 440 implements online learning methods that continuously refine model parameters as new sensor data is collected. The robot effectively “learns” from each collision or stress incident, adjusting its policies on how quickly to initiate shape reformation or how to distribute actuation power among the SMA elements. Over time, the system evolves to handle unforeseen scenarios such as novel collision angles or unexpected environmental changes with increased robustness and responsiveness.

[0068] For large-scale deployment or mission-critical applications, according to some embodiments, the central controller 402 can synchronize with a digital twin 450 running in an edge server or cloud-based platform. The digital twin 450 emulates the realtime state and conditions of the SMA structure 220, the servos/joints 202, and the environment. Al models in the cloud are configured to run high-fidelity simulations using advanced physics engines to test new deformation strategies or motion plans. If a more optimized approach is discovered, the refined policy is sent to the robot’s central compute unit 402. This cyclical process accelerates continuous improvement, ensuring the local Al module 440 is always updated with the latest best practices for shape reformation and impact mitigation. [0069] According to some embodiments in settings such as industrial automation or healthcare, human operators may oversee the robot’s Al-driven decisions via a supervisory dashboard 470. The operator can adjust parameters — e.g., maximum allowed stiffness, preferred reformation rate, or safety thresholds — based on operational context. The Al module 440, in turn, can present predictive insights and recommended courses of action (like “stiffen limb segment B now” or “reduce speed in corridor 3”) so that the operator is informed and can override or confirm the robot’s adaptation strategy when necessary.

[0070] In preferred embodiments, each SMA element 422 is interfaced with a dedicated local control channel of a microcontroller 418 on the PCB 408. This localized control channel includes both sensing and actuation lines so that parameters such as electrical current, electrical resistance, voltage drop, and temperature can be measured in real time at the SMA element 422. For example, each SMA element 422 may be connected to a pair of sensing leads that provide a continuous data stream to the microcontroller 418 or a field-programmable gate array (FPGA). In this way, the local controller 418 can accurately determine the SMA element’s current state of deformation, track internal thermal conditions, and detect anomalies like sudden force spikes or local short circuits.

[0071] Once these sensor readings are acquired, they are processed by a low- latency feedback loop within the microcontroller 418 or FPGA to maintain or adjust the SMA element’s configuration by modulating current to the SMA element 422. Typically, a proportional-integral-derivative (PID) controller is employed to compare the measured state (e.g., a certain strain or temperature reading) against a target reference (e.g., a desired angle, stiffness, or temperature range). The PID controller continuously refines the control signal — namely, the current delivered to the SMA element 422 ensuring that the actual shape of the SMA element 422 converges to the desired setpoint with minimal overshoot or oscillation. This design can be extended to multiple control loops, each responsible for a specific SMA strand of a larger SMA element 422, thereby distributing computational load and reducing latency. [0072] According to some embodiments, a more resource-intensive machine learning (ML) module (e.g., a neural network or reinforcement-learning agent) may be run on a companion processor, such as a single-board computer or edge Al module 440. The ML module ingests historical data correlating impact force profiles, strain readings, and the resulting deformation responses over time. By leveraging these datasets stored locally on solid-state memory 460 or remotely in a cloud-based database, the ML module can forecast the likelihood, location, and seventy of future impacts based on ongoing sensor readings and contextual cues (e.g., the robot’s current movement velocity or external sensor data indicating nearby obstacles). In scenarios where a collision is deemed imminent, the ML module issues pre-emptive stiffening commands to raise the structural rigidity of select SMA elements 422, thereby bracing the robot’s architecture against high-impact forces.

[0073] In addition to thermally or electrically induced shape changes, each SMA element 422 can also serve as a tactile sensor capable of generating unique electrical “signatures” in response to specific impact events. By actively monitoring minute variations in current draw across numerous strands, the system can pinpoint the location and vector of applied forces. Such resolution is achieved through multi-node comparison, wherein multiple SMA elements 422a, 422b near the impact site exhibit concurrent fluctuations in measured values. These impact signatures are then aggregated and analyzed by either a local microcontroller 418a or the central controller 402, which can interpret not only the magnitude of the force but also the directional bias (e.g., from left to right, top to bottom). This advanced sensing mechanism unlocks sophisticated haptic feedback capabilities, allowing the robot to detect subtle tactile interactions, including partial contact or glancing blows, which might otherwise go unnoticed in traditional systems.

[0074] Beyond immediate reactive control, the aggregated data including instantaneous current spikes to long-term wear patterns is logged in a robust data structure. This log may reside locally on flash memory 460, be mirrored across a distributed network of microcontrollers, or be uploaded to a secure cloud database. Predictive maintenance algorithms can then operate on the historical dataset to detect degradation trends such as increasing actuation times (indicative of changes in SMA phase transition efficiency) or erratic current consumption (indicative of damage to the strand or insulation). With these insights, the system 400 can schedule proactive repairs or component replacements before catastrophic failures occur, reducing operational downtime.

[0075] Furthermore, by implementing standardized software interfaces (e.g., RESTful APIs, MQTT topics, or industrial loT protocols like OPC UA), the system 400 can share real-time and historical performance metrics with remote monitoring stations, factory-floor supervisors, or external Al analytics platforms. This connected architecture ensures that stakeholders can monitor the robot’s health, reconfigure control parameters if needed, and continuously refine ML models for impact forecasting. In essence, the system’s adaptive control strategies, multi-tiered sensor feedback, and advanced analytics form a closed-loop ecosystem that bolsters resilience, responsiveness, and intelligence within the robotic structure.

[0076] In various embodiments, communication pathways within the control system 400 are organized around a distributed control architecture that allows local decision-making while facilitating global coordination of the SMA elements 422. This architecture typically comprises multiple microcontrollers 418a, 418b, 418n each situated at or near a servo motor, joint, or limb segment of the robot, and at least one central controller 402 (e.g., a CPU) or edge Al module 440. The following paragraphs describe the hierarchical processes by which deformation and reformation commands are generated, disseminated, and executed.

[0077] Each SMA element 422 (i.e., a single SMA strands or a localized SMA cluster/mesh within a single limb segment) is wired to a dedicated microcontroller 418 or system-on-chip (SoC). The microcontroller 418 measures local strain, current, temperature, and other parameters from the SMA elements, updating its internal state model at a high sampling rate (e.g., thousands of times per second). Upon detecting significant events (e.g., a spike in strain indicative of a collision), a given microcontroller 418a immediately sends an event flag, e.g., a small data packet containing sensor identifiers, timestamps, and measured values, over a dedicated bus 419 or network to adjacent microcontrollers 418b and/or the central controller 402. This decentralized reporting structure ensures that critical data is relayed quickly without overburdening the entire network.

[0078] Neighboring microcontrollers 418a, 418b, 418n within the same limb (or SMA cluster) may be linked by a high-speed serial bus 419 (e.g., SPI, RS-485, or a proprietary multiplexed protocol) to share real-time sensor readings. If one microcontroller 418a identifies an oncoming impact, it can alert its peer microcontrollers 418b, 418n to pre-emptively adjust their SMA elements 422 for bracing or softening. This peer-to-peer synchronization avoids the need for all decisions to route through the central controller 402 which is beneficial for time-critical collision responses. The microcontrollers 418 might employ a token-based or master-slave communication scheme to avoid data collisions on the bus 419, ensuring that each microcontroller’s warnings or status updates arrive reliably and in sequence.

[0079] At higher layers, the central controller 402 (which according to various embodiments may be an onboard processor, an industrial programmable logic controller, or an external Al gateway) continuously polls each microcontroller’s 418 status registers to form a global state map of the robot’s condition. The central controller 402 aggregates data about temperature differentials, deformation angles, and electrical consumption across all SMA elements 422, using these aggregated readings to spot system-wide trends or imminent issues such as heat buildup in a particular limb or recurring high- impact zones. When the central controller 402 detects patterns warranting broader action (e.g., resetting multiple limbs to neutral shape after a significant collision), it transmits reformation commands to each relevant microcontroller 418, specifying the target shape or stiffness level for each SMA element 422 that is modulated by that microcontroller 418.

[0080] Deformation or reformation commands include parameters for desired strain offset, activation current, ramp-up speed, and target hold duration. Upon receiving these commands, each local microcontroller 418 calculates a stepwise or continuous current profile for its assigned SMA element(s) 422 to reach the requested configuration. During execution, the microcontrollers 418 exchange periodic progress updates both with other microcontrollers 418 (for local synchronization) and with the central controller 402 (to confirm successful or partial reformation). This layered approach ensures that partial reformation is possible if certain SMA elements 422 have mechanical or electrical constraints, preventing a single fault from halting the entire shape adaptation process.

[0081] In certain embodiments, the robotic system employs real-time communication protocols such as CAN bus, Ethernet/IP, or EtherCAT to ensure deterministic data transfer with bounded latency. Reliability is bolstered by error-detection and correction features, such as cyclic redundancy checks (CRC) or Hamming codes, embedded within each data packet. Security measures, for example, message authentication codes (MACs) and encrypted tunnels, may be employed to guard against unauthorized command injection or tampering, especially important in industrial or medical settings where the risk of malicious interference or unintended control signals must be minimized.

[0082] If a bus 419 segment or a microcontroller 418 fails, the system can revert to reduced functionality modes. For instance, unaffected components can still deform or reform based on local sensor inputs, while the failed node’s state is flagged for inspection. This redundancy ensures that the robot remains at least partially operational rather than entering a complete shutdown due to a single point of failure. In some embodiments, alternative communication links such as wireless backup channels (e.g., BLE, Wi-Fi) may be activated as a fallback path for critical reformation commands.

[0083] During normal operation, an event-driven logging system records each local event flag, global command, or sensor anomaly. As these logs accumulate, they are periodically uploaded to a central server or cloud-based analytics platform, enabling advanced correlation of historical impact data with real-time shape reformation patterns. The system 400 can thus learn from past experiences, refining predictive models (e.g., anticipating collisions in specific zones) or adaptive feedback strategies (e.g., modulating actuation intensities in known high-wear areas).

[0084] In an industrial environment, the robot’s deformation and reformation processes may be coordinated with external machinery or safety systems. For example, according to some embodiments, a manufacturing execution system (MES) or a safety programmable logic controller (PLC) 480 can signal upcoming high-force interactions (like automated tool changes or part deliveries), prompting the robot to enter a protective configuration. Likewise, an operator panel or remote diagnostic interface might provide manual commands or override signals if an operator needs to place the robot into an emergency “rigid” or “soft” mode. By exposing standardized APIs or fieldbus mappings, the robot’s shape-changing abilities can be integrated seamlessly within broader factory automation, loT, or healthcare ecosystems.

[0085] Beyond typical robotic arms or manipulators, the apparatus, systems and methods described herein can be adapted to mobile robotics, including quadrupeds or humanoid robots, where impact absorption is critical for safe falls, uneven terrain navigation, or accidental collisions with people or objects. In such embodiments, the SMA protective components could be integrated into leg segments, torsos, or protective exoskeleton shells, providing both shock absorption and structural reinforcement.

[0086] The described apparatus, systems and methods are also applicable to aerospace or drone applications, where SMA protective components can be integrated into landing gear or protective housings that deform upon landing impact or turbulence, then reform mid-flight for aerodynamic efficiency. In underwater robotics, specialized SMAs resistant to corrosion can serve as pressure-tolerant skins that flex in response to water pressure at depths or to absorb underwater collisions with debris, coral, or other structures. Coupled with embedded pressure sensors, these systems can dynamically adjust buoyancy and shape for improved maneuverability.

[0087] For medical and rehabilitation devices, the described embodiments be scaled to create exoskeletons that fit around limbs, measuring muscle signals (EMG) and adjusting support or resistance accordingly. In physical therapy scenarios, these devices could measure feedback on how patients apply force and automatically adjust stiffness to guide a user’s range of motion. This synergy between adaptive mechanical support and integrated sensing capabilities opens new avenues for customized therapy, real-time motion analysis, and safer human-robot interaction.

[0088] While the above description provides examples of one or more apparatus, methods, or systems, it will be appreciated that other apparatus, methods, or systems may be within the scope of the claims as interpreted by one of skill in the art.

Claims

Claims:

1 . A shape memory protective component comprising: first and second servo motors; a tube connecting the servo motors for routing cabling between the first and second servo motors; a shape memory alloy (SMA) structure surrounding the tube and configured to deform and dissipate energy by hysteresis in response to external force applied to the SMA structure, and reform to an original shape when electric current is applied to the SMA structure; and first and second printed circuit boards (PCBs) mounted to the first and second servo motors and configured for supplying electric current to the SMA structure.

2. The shape memory protective component of claim 1 , further comprising a sensor embedded in an exterior-facing surface of the SMA structure, wherein the sensor is configured to sense input from a user.

3. The shape memory protective component of claim 1 , further comprising a sensor embedded in an exterior-facing surface of the SMA structure, wherein the sensor is configured to sense an environment around the shape memory protective element.

4. The shape memory protective component of claim 1 , wherein the SMA structure comprises a plurality of SMA elements, each SMA element comprising: a SMA stand or a SMA cluster of SMA strands.

5. The shape memory protective component of claim 4, wherein each SMA element is connected to a microcontroller on a PCB, the microcontroller configured to sense a state of the SMA element and modulate current applied to the SMA element.

6. The shape memory protective component of claim 5, wherein the microcontroller is configured to sense one or more of: a local strain, an applied current, an electrical resistance, a voltage drop and a temperature of the SMA element.

7. The shape memory protective component of claim 5, wherein the microcontroller is operably connected to an adjacent microcontroller on the PCB, the adjacent microcontroller configured to modulate current applied to an adjacent SMA element in response to a signal from the microcontroller.

8. The shape memory protective component of claim 1 , wherein the SMA structure comprises slots, openings and/or trenches filled by a secondary material configured to deform, absorb and/or dissipate energy by hysteresis in response to external force applied to the SMA structure and/or promote the SMA structure to reform to the original shape when electric current is applied to the SMA structure.

9. The shape memory protective component of claim 1 , further comprising a conductive sleeve surrounding at least a portion of the protective element, the conductive sleeve comprising a conductive fabric or a conductive mesh.

10. The shape memory protective component of claim 9, wherein the conductive sleeve is connected to at least one PCB.

11. The shape memory protective component of claim 9, wherein the conductive sleeve is contiguous with the SMA structure.

12. The shape memory protective component of claim 9, further comprising a sensor embedded in an exterior-facing surface of the conductive sleeve, wherein the sensor is configured for sensing input from a user.

13. The shape memory protective component of claim 12, further comprising a sensor embedded in an exterior-facing surface of the conductive sleeve, wherein the sensor is configured to sense the environment around the shape memory protective element.

14. The shape memory protective component of claim 1 , wherein the protective element forms at least part of a limb segment of a robot apparatus.

15. The shape memory component element of claim 1 , wherein the first and the second servo motors are disposed at respective first and second joints of a robot apparatus.

16. A robotic system comprising: at least one limb segment comprising a SMA structure configured to deform and dissipate energy by hysteresis in response to external force applied to the SMA structure, and reform to an original shape when electric current is applied to the SMA structure; a plurality of sensors configured for sensing an environment around at least the limb segment; and a control system configured to: receive input from the plurality of sensors to detect a potential collision between the limb segment and an object in the environment; and modulate current applied to the SMA structure in response to detecting the potential collision.

17. The robotic system of claim 16, wherein the SMA structure comprises a plurality of SMA elements, each SMA element comprising: a SMA stand or a SMA cluster of localized SMA strands, wherein the control system comprises: a plurality of microcontrollers, each microcontroller configured to: sense a state of an SMA element; and modulate current applied to the SMA element; a central controller configured to: receive input from the plurality of sensors to detect the potential collision between the limb segment and an object in the environment; and signal one or more microcontrollers to modulate the current applied to one or more SMA elements.

18. The robotic system of claim 17, wherein the central controller is further configured to implement a predictive model trained to recognize precursor events to a collision and preemptively signal the one or more microcontrollers to modulate the current applied to the one or more SMA elements.

19. A method for vibration and impact mitigation comprising: providing an apparatus composing a SMA structure configured to deform and dissipate energy by hysteresis in response to external force applied to the SMA structure, and reform to an original shape when electric current is applied to the SMA structure; constantly applying the electric current to the SMA structure to substantially maintain the SMA structure in the original shape; and performing an activity which causes the apparatus to experience vibrations and impacts.

20. The method of claim 19, wherein the apparatus is a robot apparatus.