Force-sensing sensor arrangement and method for manufacturing the sensor arrangement and components of the sensor arrangement - Patents.com

Abstract

Improve sensor configurations for robotics. The invention relates to a sensor arrangement for detecting forces, the sensor arrangement comprising a measuring surface and an optical detector for detecting reflected light. The invention further relates to a corresponding manufacturing method.

Description

[Technical Field]

[0001] The present invention relates to a force-sensing sensor arrangement (sensor device), a method for making the top of a force-sensing sensor arrangement, and a method for making a sensor arrangement. [Background technology]

[0002] When developing applications such as robotics, sensing forces applied to the robot's hands or other parts of the robot, such as legs or manipulators, is important to enhance the robot's ability to move around and manipulate objects. Known implementations of sensor configurations that can be used in robotic applications to get feedback on applied forces are very expensive and do not have sufficient resolution. Summary of the Invention [Problem to be solved by the invention]

[0003] It is therefore an object of the present invention to provide a force-sensing sensor arrangement that is different or optimized with respect to the prior art. It is a further object to provide a corresponding method for producing such a sensor arrangement or a component part thereof. [Means for solving the problem]

[0004] The object is achieved by the subject matter of the main claim. Preferred embodiments can be derived, for example, from the dependent claims. The subject matter of the claims is made the subject matter of the description by explicit reference.

[0005] The present invention relates to a force-sensing sensor arrangement. The sensor arrangement includes a base. The sensor arrangement includes a top having an elastically deformable wall. The top is attached to the base such that the top and base define an interior space. The wall includes an outer measurement surface and an inner reflective surface, where the reflective surface partially bounds the interior space.

[0006] The sensor arrangement comprises a light source arrangement mounted to the base and comprising a plurality of light sources arranged to emit light towards the reflective surface, and the sensor arrangement comprises an image sensor having a detection surface that views at least a portion of the reflective surface.

[0007] Using such a sensor configuration, forces applied to a measurement surface can be detected by optical image recognition. Light reflected from the reflective surface generates a pattern on the image sensor that depends in a highly sensitive way on the force applied to the measurement surface. Therefore, such light patterns can be used to perform highly sensitive force inferences.

[0008] In particular, the light source may generate structured light within the interior space.

[0009] The base is typically the component that defines the base of the sensor arrangement. For example, it can be fixed to some holding means, especially a robotic component. The top is typically the component that contacts the object to which the force is applied, or that applies the force to the sensor arrangement.

[0010] When an element is attached to a base, it may mean that it is attached to another element of the base and / or that it is part of the base. This may be the case for elements such as a light source arrangement, a collimator, or an image sensor. For example, an element attached to a base may be attached to a support structure, which may provide, for example, stability or connectivity to other elements.

[0011] A wall is typically a component that comes into contact with an external object and deforms when the object applies a force, or vice versa. Possible implementations of walls are described further below.

[0012] The interior space is typically completely surrounded and / or defined by the top and base. The interior space may be hollow; however, it may be filled with a substance. Typically, light from the light source arrangement propagates through the interior space toward a reflective surface, where it is reflected and propagates to the image sensor.

[0013] The term "sensor arrangement" is to be understood in particular as an arrangement of several components having the function of a sensor, in particular for force detection.

[0014] In particular, each light source may have its own color. The light source arrangement preferably comprises light sources with at least two or three different colors. This results in different colored light patterns on the detection surface, thus allowing better evaluation possibilities, since the light can be identified as originating from one or at least some of the light sources. However, the same color may also be used for the light sources.

[0015] The light source and the detection surface are preferably arranged such that light emitted by the light source and reflected by the reflective surface generates a light pattern on the detection surface. This light pattern can be used for evaluation purposes, in particular using neural networks. The light pattern has been shown to be very sensitive to the force applied to the measurement surface. This is also related to the direction and shape of the indenter force. It has therefore been found that such a light pattern constitutes a strong indicator of the applied force.

[0016] In particular, the light pattern on the sensing surface changes with deformation of the measurement surface, and such changes can be evaluated to infer forces or force maps.

[0017] In particular, the color distribution of light reflected from a reflecting surface onto a sensing surface changes as the measurement surface deforms. This color distribution can change especially when different colors are used as light sources. The change in color distribution has been shown to be very sensitive to applied force.

[0018] In one implementation, the reflective surface is diffusely reflective. This may mean, among other things, that light striking the reflective surface is not reflected as in a mirror, but rather as in an at least slightly rough surface. Light may be slightly attenuated upon reflection from a diffusely reflective surface. In an alternative implementation, a specularly reflective surface may also be used.

[0019] In particular, the image sensor can be mounted in the base and / or the interior space, which is a simple and reliable implementation and reduces the distance to the reflective surface, although the image sensor may be mounted elsewhere.

[0020] One, some or all of the light sources may have an adjustable color. This is particularly useful for setting the color during a setup procedure. However, fixed colors can also be used.

[0021] One, some or all of the light sources may have adjustable brightness. This is particularly useful for setting the brightness during the setup procedure. However, fixed brightness can also be used.

[0022] Preferably, the light sources are arranged around the image sensor, which may lead to a favorable distribution of light within the interior space.

[0023] Preferably, the light source is positioned to emit light in a way that results in a distribution of reflected light on the detection surface that does not result in an intensity that exceeds the saturation density. This prevents oversaturation, which can lead to errors in force estimation. For example, variables that can be adjusted to prevent oversaturation are the brightness of the light source, the position and size of the collimator, the camera exposure time, and the lens. However, other variables can also be used for this purpose.

[0024] According to a preferred embodiment, the light source is a light emitting diode. Such a light emitting diode is a reliable light source, however, other types can also be used.

[0025] Preferably, the sensor arrangement comprises a plurality of collimators, each collimator being assigned to one light source and defining a respective illumination angle and / or a cone of emitted light, which allows for very precise control of the propagation of light rays onto the reflective surface.

[0026] One, some, or all collimators may be positioned off-center (acentric) relative to their assigned light source, which provides suitable light distribution for many applications, however, a centered position may also be used.

[0027] The collimator can in particular be embodied as a hole in a collimator ring, which represents an easy realization of the collimator, and the collimator ring may block light other than that passing through the hole.

[0028] The light source and / or collimator can be arranged to cover with light at least 80%, at least 85%, at least 90%, or 100% of the reflecting surface or the intended measurement area or the area inside the measurement surface or the intended measurement surface. Thus, a large portion of such surface or area can be used. The intended measurement area can be, in particular, a certain area, i.e., a subpart of the measurement surface. The inner area of ​​the measurement surface can be an area on the wall opposite the measurement surface. The intended measurement surface can also be a subpart of the measurement surface, where the measurement surface can in principle be defined as a surface that can be used for force inference.

[0029] The light sources and / or collimators can be positioned so that at least 60%, at least 70%, at least 80%, at least 90%, or 100% of the reflective surface is directly illuminated by light from up to four light sources and / or light from at least two light sources. This produces a light pattern on the detection surface that has been proven to be usable for significantly better power inference. A directly illuminated component is specifically one in which light rays propagate directly from the light source to the component without being reflected from another component of the reflective surface.

[0030] In one implementation, the collimator has a collimator hole diameter of at least 0.8 mm. In another implementation, the collimator has a collimator hole diameter of up to 4 mm. Such diameters have proven suitable for typical applications. However, diameters and shapes other than round holes can also be used.

[0031] The light sources may be arranged in a circular pattern at the base, so that they form or define a circle, which provides a predictable light pattern and is particularly suitable for typical implementations where the top is circularly symmetric.

[0032] The detection surface may be constructed and arranged to view at least 70%, at least 80%, at least 90%, or even the entire reflective surface, which may mean, among other things, that the detection surface is capable of detecting light reflected from such a portion of the reflective surface, thereby providing high measurement accuracy.

[0033] In particular, the image sensor may be a color camera sensor or may be color sensitive, which in principle allows the image sensor to be used with enhanced measurement capabilities compared to black and white or grayscale sensors.

[0034] In particular, the image sensor may comprise a plurality of pixels, each configured to individually detect light, and thus the pixels may determine the resolution of the photodetector.

[0035] The detection surface may face the reflective surface, so light reflected from the reflective surface may propagate directly to the detection surface and ultimately through some optical means, such as a wide-angle lens as described herein, although other implementations are possible.

[0036] The sensing surface may be parallel to the inner surface of the base, which allows for simple implementation, although other orientations are possible.

[0037] The detection surface may be configured to detect light patterns and / or images, which may be used for force inference, as described herein.

[0038] The image sensor may be configured to detect light patterns and / or images at a frame rate, whereby frames are typically detected successively with a fixed time interval between them.

[0039] In particular, the frame rate may be at least 10 fps (frames per second), at least 20 fps, at least 30 fps, at least 50 fps, or at least 100 fps. In particular, the frame rate may be up to 30 fps, up to 50 fps, up to 100 fps, or up to 200 fps. Such values ​​have proven suitable for general use. Each lower value can be combined with a higher value to form a suitable interval. However, other values ​​may also be used.

[0040] The image sensor may include a wide-angle lens or a fisheye lens optically positioned between the reflective surface and the detection surface, which may improve image detection, for example, the image detector may ensure observation of a specific portion of the reflective surface.

[0041] In particular, the interior space may be a hollow space, which may mean that the space is filled with air, but it can also be filled with another substance, in particular an optically transparent substance such as a fluid, a solid, glass, or an elastomer.

[0042] At least one of the light source and the collimator may be configured to emit light in respective cones. Such cones may be defined by respective cone angles that delimit the outer dimensions of the cones. The cones of light may have a cross-sectional area that steadily increases with increasing distance from the light source. The collimator may also be used to determine such a shape of the light after it passes through the collimator.

[0043] In particular, one, some or all of the cones may have cone axes that are inclined outward by more than 0° relative to at least one of the common axis and an axis perpendicular to the base. This may lead to the prevention of oversaturation at the tip of the apex. This may improve the measurability of forces near the tip. The base may in particular extend along a plane to which the axis may be perpendicular, i.e., vertical.

[0044] In particular, one, some or all of the cones may have cone axes inclined outwards by up to 10° relative to a common axis and / or an axis perpendicular to the base, which has proven suitable for general applications, although larger angles may also be used.

[0045] In particular, one, some or all of the cones may have an outer cone angle of at least 35° and / or at most 80°. The outer cone angle may in particular be the angle of the outer boundary of the cone relative to the cone axis and / or the central propagation direction. The cone axis may define the center of the cone and / or the light propagation direction.

[0046] In particular, the cones may partially overlap in a plane perpendicular to the common axis, where the overlap depends on the distance between the plane and the base. This will result in portions of the reflective surface that are not illuminated being omitted and therefore unavailable for force inference. In particular, the light source and image sensor may be arranged such that light emitted by the light source and reflected by the reflective surface is detected by the image sensor.

[0047] According to one implementation, the reflective surface is covered with a pattern and / or a plurality of trackable objects, which may improve the accuracy of force inference. According to an alternative implementation, the reflective surface is a smooth surface.

[0048] According to one embodiment, the sensor configuration has a common axis. In particular, the top may be completely circular about the common axis. This corresponds to a simple implementation. Alternatively, the top may be partially circular about the common axis. This may allow for specific configurations, such as thin sections as described herein.

[0049] The detection surface may in particular be perpendicular to the common axis.

[0050] The light sources may be arranged to emit light parallel to a common axis, which may result in a geometrically simple implementation, in particular before possible deflection or obstruction by a collimator.

[0051] In particular, different parts of the reflective surface may be illuminated by light sources of different colors from different directions, thereby improving the power of force inference.

[0052] According to one embodiment, the tip can be tapered with an outer diameter that decreases with increasing distance from the base, which would provide a profile suitable for typical use cases where the tip is used for manipulation purposes or specific force measurements.

[0053] In particular, the top and / or the wall may be conical in shape.

[0054] In particular, the wall may be configured to relay deformations from the measurement surface to the reflective surface. This may particularly mean that a force applied to the measurement surface will induce a deformation of the measurement surface, which deformation will be relayed through an inner part of the wall to the reflective surface, which will also undergo a deformation resulting in a different reflection of light at a particular point or area. This different reflection will result in a different light pattern on the detection surface, which may be measured and evaluated.

[0055] Preferably, the top is removably attached to the base, allowing the base to be reused with a different top, for example, if the top is damaged or loses properties relevant for metrology purposes. For example, the base may remain fixed to part of a robot or other component, and the top may be replaced as needed.

[0056] Removable attachment may mean, inter alia, that some means are provided for releasing the top from the base without damaging the components.

[0057] In particular, the top may be attached to the base in a manner that allows the top to be interchangeable.

[0058] The top may be releasably attached to the base by a bayonet attachment and / or a threaded connection, additionally or alternatively by a pair of tangs and corresponding notches. Such connections have been proven to provide easy interchangeability and a secure connection. However, other connection schemes may also be used.

[0059] In one implementation, the top comprises only walls, which typically provide sufficient stability by themselves. In particular, the walls may comprise only a homogenous wall material, especially without a supporting structure made of another material.

[0060] According to one implementation, the top comprises a body located inside the wall. Such a body may provide additional rigidity and may be made of a different material than the wall. In particular, the body may comprise a plurality of wires or wire-like elements.

[0061] Preferably, the body is lattice-like. This may mean that the body is made of relatively thin wires or other elements that leave spaces between them. This space is typically filled with wall material. Furthermore, the body is typically completely or partially surrounded by wall material.

[0062] The body may preferably be made of steel, stainless steel, or aluminum. Such materials provide sufficient stability. However, other materials may also be used. In particular, any material that can withstand the desired maximum force and / or can be manufactured into the required shape (e.g., by three-dimensional molding) may be used.

[0063] For example, aluminum or stainless steel can be used for the body, although other materials such as copper, bronze, brass, carbon fiber, etc. are also possible.

[0064] In particular, the body is rigid or semi-rigid, which provides sufficient rigidity.

[0065] In particular, the wall may comprise a wall material, which may be suitable for transmitting deformations from the measurement surface to the reflecting surface.

[0066] Preferably, the wall material comprises an elastomer, such as Smooth-On Ecoflex 00-30, Ecoflex 00-35, or Ecoflex 00-50. Soft elastomers with particularly high elongation can be used. For example, the wall material can have an elongation of at least 800% and / or up to 1,000% or 900%.

[0067] Preferably, the wall material contains aluminum powder and / or aluminum flakes, which provide a desirable reflectivity for the reflective surface. For example, the powder and / or flakes are present to create a suitable type of reflective surface. Deformation or changes in angle relative to the light source can cause changes in shading. In particular, aluminum powder can be used to deflect ambient light, while aluminum flakes can be used to enhance reflectivity. The powder typically has a smaller diameter than the flakes.

[0068] The body is preferably surrounded by the wall material, which improves the stability of the wall material.

[0069] In particular, the body may be completely overmolded with the wall material, or alternatively, the body may be only partially overmolded with the wall material.

[0070] The skeleton may be particularly reusable. This may mean, for example, that the wall material surrounding the skeleton can be dissolved in some solvent, leaving the skeleton without the surrounding wall material. New wall material can then be overmolded.

[0071] In particular, the wall thickness may be at least 0.8 mm or at least 1.2 mm.

[0072] In particular, the wall thickness is at most 4 mm or at most 5 mm.

[0073] While such values ​​are suitable for general use, other values ​​may also be used. The values ​​given may relate to portions of the wall outside particularly thin areas.

[0074] According to one implementation, the wall comprises a thinned area having a smaller thickness than the rest of the wall, which will provide a particularly high sensitivity in this localized area.

[0075] The thinned area may be located especially on the side opposite the base, where high sensitivity may be preferred.

[0076] The thin area may be shaped according to a fingernail (area). For example, the sensor configuration may be shaped like a thumb or another finger. The thin area may be placed at the location of a fingernail (area). The thin area may be completely or at least substantially flat, or may only be slightly curved. In general, the thin area may have any shape.

[0077] In particular, the thin area may have less than one-quarter of the measuring surface, thereby providing an appropriate increased measuring capability without compromising stability.

[0078] Preferably, the thinned area has a thickness of at most 0.8 mm or at most 1.2 mm and / or a thickness of at least 30% or at most 50% of the thickness of the wall outside the thinned area. This provides a suitable increase in measurement sensitivity. However, other values ​​can also be used. The wall may have a uniform thickness outside the thinned area. However, if the wall thickness is not uniform, the criterion for defining the relative thickness of the thinned area may be an average value.

[0079] The sensor arrangement can be a fingertip and / or a robotic manipulation element. The sensor arrangement can therefore have two functions: to manipulate the element and to measure the applied force. However, other implementations are possible.

[0080] The invention further relates to a method for making a top for a sensor construction, the method comprising the following steps: - The process of providing a structure made of structural materials, which encloses an interior space. - covering the body with a wall material such that the wall material forms an elastically deformable wall defining an outer measurement surface and an inner reflective surface, wherein the reflective surface bounds an interior space.

[0081] This allows the top to be manufactured simply and reliably. For details of the elements, reference is made to the descriptions given elsewhere in this document. All statements given regarding the method aspects are in principle applicable to the structural aspects and vice versa.

[0082] In particular, the body material may be stronger than the wall material, thus providing stability. The body may be more rigid than the wall material.

[0083] The step of providing a body may comprise three-dimensional modeling of the body. In other words, the body may be three-dimensionally shaped. This allows for greater flexibility in designing the body. However, other processes may also be used.

[0084] The body may be made of a plurality of wires with openings formed therebetween, and such openings may be filled with a wall material.

[0085] Preferably, the body may be in the shape of a dome or a cone.

[0086] The covering can in particular be made by overmolding, which is a reliable method of placing a material such as an elastomer around a body, although other methods can also be used.

[0087] The interior space may be hollow inside the top, but may also be filled with, for example, an optically transparent material.

[0088] The method may further comprise covering the reflective surface with a pattern and / or a number of trackable objects. Alternatively, the reflective surface may remain smooth.

[0089] In particular, structured casts may be applied to generate patterns, and such structured casts may correspond to patterns.

[0090] The bodies may have a common axis, and the bodies may be partially or completely circular about the common axis.

[0091] The body may have an outer diameter that tapers toward the tip, which may correspond to the preferred shape of the tip.

[0092] The wall material may be configured to relay deformation from the measurement surface to the reflecting surface.

[0093] The body may be in the form of a lattice.

[0094] The body may be made of steel, stainless steel or aluminum. For further alternatives, reference is made to the descriptions given elsewhere in this specification.

[0095] The body may be rigid or semi-rigid.

[0096] The wall material may comprise an elastomer.

[0097] The wall material may include aluminum powder and / or aluminum flakes.

[0098] The body may be covered with a wall material such that the wall material surrounds the body.

[0099] The wall material is formed to a thickness of at least 0.8 mm or at least 1.2 mm and / or formed to have a thickness of at most 4 mm or at most 5 mm.

[0100] The wall material may be formed with thinned areas that have a thickness less than other portions of the wall material (portions other than the thinned areas), which may be particularly useful for light touch detection and shape discrimination.

[0101] The thinned area may be located near the tip of the body.

[0102] The thin area may be shaped according to the fingernail (area).

[0103] The thin area may include less than one-quarter of the measurement surface.

[0104] The thickness of the thinned area may be at most 0.8 mm or at most 1.2 mm and / or at most 30% or at most 50% thicker.

[0105] The body may be covered with a wall material that is removably attached to the body.

[0106] The invention further relates to a method for producing a sensor arrangement, the method comprising the following steps: providing a base; creating a top portion having an interior space; attaching a light source arrangement comprising a plurality of light sources to a base; attaching an image sensor to a base; covering the base with a top such that the light source is positioned to emit light toward the interior space and the image sensor is positioned in the interior space; is.

[0107] This method provides reliable manufacturing of the sensor configuration.

[0108] In particular, the top may be made as disclosed herein, all implementations and variations being applicable.

[0109] The light source may be mounted so as to surround the image sensor.

[0110] The method may further comprise disposing a plurality of collimators over the light source, the collimators defining at least one of an illumination angle of the emitted light and a cone of the emitted light. For example, such collimators may be formed within a collimator ring, thereby effectively disposing a collimator ring.

[0111] The sensor configurations produced may be specifically embodied as described herein, and all embodiments and variations are applicable.

[0112] The invention further relates to a sensor arrangement as disclosed herein or manufactured as disclosed herein, further comprising an electronic control module configured to perform the force inference method of the sensor arrangement.

[0113] Thus, the sensor arrangement may have its own control module. For example, the control module may be an electronic entity inside the base or located near another part of the base. Alternatively, the control module may be located remotely from the base and top, and may be, for example, a computer.

[0114] The control module may be configured to execute a force inference method that provides a force map of the measurement surface and a force map comprising a plurality of force vectors. Regarding the force inference method, reference is made to the description given elsewhere in this specification. All embodiments and variants are applicable.

[0115] In particular, the force map is mm 2 (1×10 -6 m 2 ) force vector of at least 0.25 mm 2 At least 0.5 force vector per mm 2 At least 0.75 force vectors per mm 2 At least 1 force vector per mm 2 At least 1.5 force vectors per mm 2 and providing at least two force vectors per mm 2 Maximum 0.25 force vector per mm 2 Maximum 0.5 force vector per mm 2 Maximum 0.75 force vector per mm 2 Maximum 1 force vector per mm 2 Maximum of 1.5 force vectors per mm, or 2 With up to two force vectors per Alternatively, other values ​​may be used.

[0116] In particular, each force vector comprises a normal force component, a first shear force component and a second shear force component.

[0117] In particular, the first shear force component may correspond to the first shear force and the second shear force component may correspond to the second shear force, the first shear force being perpendicular to the second shear force.

[0118] In particular, the sensor configurations disclosed herein may be finger-shaped, may be soft sensors, and / or have full-circumference force sensing capabilities, may be enabled by machine learning, and may be accurate, sensitive, durable, and affordable.

[0119] There are two main techniques for obtaining three-dimensional (3D) information from a single camera that can be suitably used with the disclosed sensor configuration. Photometric stereo (PS) techniques use multiple images of the same scene with different distributed light sources to infer 3D shape from shading information. Structured Light (SL) technology is a single-shot 3D surface reconstruction technique that uses unique light patterns and the fact that they are projected differently onto 3D surfaces.

[0120] In general, SL is typically used for global reconstruction, while PS has strengths in capturing local details. Insight combines PS and SL to perform deformation reconstruction of a full 3D dome-shaped surface in a single-camera, single-image setup. Next to the camera or image detector are several light sources that generate cones of light. When an area of ​​the measurement surface is contacted and deformed, the contact area moves, causing a visible color change due to two effects: color differences due to shading, and movement between color areas of different light intensities due to the light cone.

[0121] The top and / or base may be specifically designed to prevent ambient light from entering the interior space, thereby preventing such ambient light from distorting the measurements.

[0122] Further inventive aspects are described below, which may be combined alone or in combination with other features disclosed herein, and which may also be considered separate inventive aspects and form the subject of claims.

[0123] The present invention relates to a method for force inference in a force-sensing sensor arrangement.

[0124] The sensor configuration is, for example, a sensor configuration to which the method can be applied, and in particular comprises at least: an elastically deformable wall, the wall having an outer measurement surface and an inner reflective surface, wherein the reflective surface partially bounds an interior space; a light source arrangement comprising a plurality of light sources configured to emit light toward an interior space; Image sensor installed inside the interior space is.

[0125] Regarding the configuration of the sensor, reference is made to the descriptions elsewhere in this specification. All embodiments and variants may be applied.

[0126] A method for inferring forces comprises the following steps. reading image data from the image sensor; calculating a force map on the measurement surface based on the image data, preferably using a feedforward neural network, the force map comprising a plurality of force vectors; is.

[0127] Such a method provides highly accurate force inference based on image detection. The use of a feedforward neural network eliminates the need for implementing analytical force estimation. The neural network can be trained specifically as disclosed herein. The training has been shown to be capable of detecting multiple indenters and identifying indenter location, force direction, and indenter shape, leading to highly accurate and fine-grained force inference.

[0128] The force map may in particular be a map defined on the actual measurement surface, where the force map may comprise a number of map points. At each map point, some information may be determined, for example a force vector, as will be explained further below. The force map typically provides information about the forces applied to the measurement surface. For example, such forces originate from one or more indenters pressing on the measurement surface, or from the object currently being manipulated by the sensor arrangement (for example, if the sensor arrangement is a robot fingertip).

[0129] A feedforward neural network may in particular be an artificial neural network that takes image data as input and delivers a force map as output. In principle, a feedforward neural network is an artificial neural network in which the connections between nodes do not form cycles.

[0130] In the following, the training aspect of the network is described. The training procedure referred to in this description should be considered as a step performed before the actual force measurement force inference is performed. Thus, the force inference method can be considered as a combination of a training step performed before the force inference and the force inference using the trained network. The force inference method can also be considered as the force inference itself, using a network trained accordingly. Further below, another training method is described. This training method may be performed independently of any force inference. Typically, force inference, in which an image sensor is read out and a force map is generated, is considered as an action performed in a certain use case (i.e., when the sensor configuration is used to measure or evaluate forces acting on a measurement surface). The case in which the sensor configuration is used to measure or evaluate forces acting on a measurement surface can be, for example, when the sensor configuration is currently manipulating an object or is otherwise in contact with an object that exerts pressure on the measurement surface.

[0131] According to one preferred implementation, a feedforward neural network was trained with the following steps, which were performed before force inference. - performing a plurality of force tests on the sensor configuration, each force test comprising: applying a force with an indenter to a location on a measurement surface of the sensor arrangement and simultaneously measuring the force applied by the indenter; performing multiple force tests, including simultaneously reading out image data from an image sensor; - for each force test, performing a corresponding simulation test on a model of the sensor configuration, Each simulated test comprises applying a simulated force to a simulated measurement surface of the model, thereby calculating a map of the simulated force on the simulated measurement surface; the simulated force map comprises a plurality of simulated force vectors; performing a simulated test, wherein simulated forces correspond to measured forces and are applied at locations on the simulated measurement surface corresponding to locations on the measurement surface; - training a feedforward neural network using the image data and the corresponding computed simulated force maps; is.

[0132] Such a training step can provide adequate training for a feedforward neural network, which can learn actual forces and corresponding force maps, the former obtained from measurements and the latter from simulations.

[0133] The force measurements and image data readout are typically performed, preferably at rest, while the forces are actually applied, which are then used in the simulation test.

[0134] Note that all terms denoted as "simulated" (tested) typically relate to the simulated test. For example, a simulated measurement surface is a measurement surface that exists only in the simulated test. The model can calculate a force map in a deterministic way from the applied simulated forces. For example, a simple spatial distribution around a point of force using Hertzian contact theory can be used. Alternatively, a finite element model can be used.

[0135] The simulated force may be the same as the measured force, which may mean that the simulated force has the same components in three dimensions and / or that it has the same direction and magnitude. However, the simulated force may also correspond to the measured force by a predefined relationship.

[0136] Preferably, the force tests for training the feed-forward neural network are performed using multiple indenters, each with a different indenter shape. In particular, the indenter shapes may be different. Thus, the feed-forward neural network can be trained to distinguish between different indenter shapes, i.e., to generate different force maps when different indenters are applied.

[0137] For example, the indenter shape may be selected from the group comprising at least a point, a circle, a triangular cross section, a square cross section, a hemisphere, a cube, and a cylinder. All of the above indenter shapes may be used in the training process, or only a subset may be used. Other indenter shapes may also be used.

[0138] Preferably, the simulated tests are performed with simulated forces based on simulated indenters, each having a simulated indenter shape that corresponds to the actual indenter shape used in the corresponding force test. A plurality of such simulated indenters may each have a simulated indenter shape that corresponds to the actual indenter shape used in the corresponding force test. Thus, the simulated forces applied to the simulated measurement surface more closely represent the actual forces due to the similar indenter shapes. This improves the training of the feedforward neural network.

[0139] Preferably, the feedforward propagation neural network was trained using multiple different indenter shapes, which allows the feedforward propagation neural network to be trained to distinguish between the forces generated by the different indenter shapes.

[0140] In one implementation, a feedforward neural network was trained using several different sizes of indenters, which allows the network to be trained to distinguish between the forces generated by different indenter sizes.

[0141] Preferably, the feedforward propagation neural network is trained with the indenter applied at each shear force for at least a portion of the force tests training the feedforward propagation neural network. This allows the feedforward propagation neural network to be trained to distinguish between different shear forces applied to the measurement surface. For example, the force map may comprise the simulated shear forces. In particular, different shear forces may result in different force maps.

[0142] Preferably, each measured force comprises a normal force component, a first shear force component and a second shear force component, which defines the force magnitude and direction in a coordinate system, in particular a global coordinate system, although other force representations may also be used.

[0143] Preferably, of the measured forces, the first shear force component corresponds to the first shear force and the second shear force component corresponds to the second shear force, in particular the first shear force is perpendicular to the second shear force.

[0144] Each measured force may have three components in a reference coordinate system, which may be a global coordinate system. Different representations that require only standard mathematical transformations are considered equivalent.

[0145] Preferably, the feedforward neural network is trained using multiple forces with different shear force components, which allows the feedforward neural network to be specifically trained to distinguish between different shear forces, especially since different shear forces of the actual applied forces may result in different force maps.

[0146] Preferably, the feedforward neural network is trained using multiple forces with different normal force components, so that the feedforward neural network can be trained to distinguish between various normal forces, where the normal force component may be the component of the force that is locally normal to the surface.

[0147] The force may be measured using a force sensor located in or next to the indenter, which allows for direct force measurement. In particular, the force sensor can measure not only the absolute value of the force, but also the corresponding orientation. From the orientation, the shear force can be derived.

[0148] Preferably, each simulated force vector comprises a normal force component, a first shear force component and a second shear force component, such that the simulated force vectors may resemble shear force components on a simulated force map that can be used to train a feedforward neural network.

[0149] Preferably, the first shear force component of the simulated force vector corresponds to the first shear force and the second shear force component corresponds to the second shear force, particularly the first shear force being perpendicular to the second shear force.

[0150] Preferably, each simulated force vector has three components in a reference coordinate system, which may be used to represent all forces within the method, although other representations may also be used.

[0151] Preferably, the image data on which the calculated force map is based comprises several invariant images, preferably three invariant images, in addition to the image data read out from the image sensor. The image data read out from the image sensor may be denoted as variable images. The invariant images can be set as follows: This has been shown to improve the results of feedforward propagation neural networks. The invariant images may be used without modification in all training steps and force inference steps.

[0152] Preferably, the invariant image is at least one of a grayscale gradient image, an image of the body, and a reference light pattern, as such images have proven suitable for typical force inference applications.

[0153] Preferably, the variable image as part of the image data was taken immediately before calculating the force map, so that the force map corresponds to the actual state of the sensor configuration.

[0154] In the preferred implementation, the force map is 2 (1×10 -6 m 2 ) force vector of at least 0.25 mm 2 At least 0.5 force vector per mm 2 At least 0.75 force vectors per mm 2 At least 1 force vector per mm 2 At least 1.5 force vectors per mm 2 With at least two force vectors per In the preferred implementation, the force map is 2 Maximum 0.25 force vector per mm 2 Maximum 0.5 force vector per mm 2 Maximum 0.75 force vector per mm 2 Maximum of 1 force vector per mm 2 Maximum of 1.5 force vectors per mm 2 Each force vector has a maximum of two. Lower values ​​can be combined with higher values ​​to create suitable spacing, however other values ​​can also be used.

[0155] Preferably, each force vector comprises a normal force component, a first shear force component and a second shear force component, so that the force vector may provide not only a normal force but also a shear force component.

[0156] Preferably, the first shear force component corresponds to the first shear force and the second shear force component corresponds to the second shear force. The first shear force may in particular be perpendicular to the second shear force.

[0157] Preferably, each force vector has three components in the reference coordinate system, which also indicate the direction of the force and therefore the shear force. Note that force vectors that are not normal to the local surface typically have shear forces.

[0158] Preferably, the feedforward neural network is trained or the force map is calculated using additional images of the reflective surfaces of the sensor arrangement without external impact as part of the image data, which may improve detection accuracy.

[0159] Preferably, an image of the wall structure of the sensor configuration is used as part of the image data to train the feed-forward neural network or calculate the force map, which may also improve detection accuracy.

[0160] Preferably, a grayscale gradient image for position encoding is used as part of the image data to train a feedforward neural network or calculate a force map, which may also improve detection accuracy.

[0161] Such additional images, e.g., images of the reflective surface without external impact, images of the body, and images of the grayscale gradient, may increase detection accuracy and / or provide better training. These images may be invariant images or may be used as part of the image data in addition to the variable images that may be read out from the image sensor. The images of the reflective surface without external impact may in particular be images of the reflective surface taken when no force is applied to the measurement surface.

[0162] In the following, we describe an alternative method for training a feedforward neural network, which is not part of the force inference method but is performed separately from the network training. For each feature, reference is made to the previous descriptions of the network training and force inference methods to avoid repetition.

[0163] The present invention relates to a method for training a feedforward propagation neural network, The feedforward propagation neural network preferably calculates a force map on a measurement surface of the sensor arrangement based on image data of the image sensor, the force map comprising a plurality of force vectors; Here, the feedforward neural network is trained in the following steps: - performing a plurality of force tests on the sensor arrangement, performing a plurality of force tests, each force test comprising applying a force with one indenter to a location on the measurement surface of the sensor arrangement, simultaneously measuring the forces applied by the indenters, and simultaneously reading out image data from the image sensor; - for each force test, performing a corresponding simulation test with one model of the sensor configuration, Each simulated test comprises applying a simulated force to a simulated measurement surface of the model, thereby calculating a map of the simulated force on the simulated measurement surface; the simulated force map comprises a plurality of simulated force vectors; a simulated force corresponding to the measured force and applied at a location on the simulated measurement surface corresponding to a location on the measurement surface; running a mock test; - training a feedforward propagation neural network using the image data and the corresponding simulated force maps; is.

[0164] In one implementation, the power test to train the feedforward neural network is performed using multiple indenters, each with a respective indenter shape.

[0165] In one implementation, the indenter shape is selected from the group comprising at least a tip, a circle, a triangular cross section, a square cross section, a hemisphere, a cube, and a cylinder.

[0166] In one implementation, the simulated test is performed with multiple simulated forces applied with or based on multiple simulated indenters, each preferably having a simulated indenter shape corresponding to the actual indenter shape used in the corresponding force test.

[0167] In one implementation, simulated tests are performed at multiple simulated forces based on multiple simulated indenters, each with a simulated indenter shape corresponding to the actual indenter shape used in the corresponding force test.

[0168] In one implementation, a feedforward neural network is trained using multiple indenters with different sizes.

[0169] In one implementation, the feedforward propagation neural network is trained with multiple indenters having respective shear forces for at least a portion of the force trials that train the feedforward propagation neural network.

[0170] In one implementation, each measured force comprises a normal force component, a first shear force component, and a second shear force component.

[0171] In one implementation, among the measured forces, a first shear force component may correspond to the first shear force, a second shear force component may correspond to the second shear force, and the first shear force may be perpendicular to the second shear force.

[0172] In one implementation, each measured force has three components in the reference coordinate system.

[0173] In one implementation, a feedforward neural network is trained using multiple forces with different shear force components.

[0174] In one implementation, a feedforward neural network is trained using multiple forces with different normal force components.

[0175] In one implementation, the force is measured using a force sensor placed in or next to the indenter.

[0176] In one implementation, each vector of simulated forces has a normal force component, a first shear force component, and a second shear force component.

[0177] In one implementation, of the simulated force vectors, a first shear force component corresponds to the first shear force, a second shear force component corresponds to the second shear force, and the first shear force may be specifically perpendicular to the second shear force.

[0178] In one implementation, each simulated force vector comprises three components in the reference coordinate system.

[0179] According to one implementation, a feedforward neural network is used in the force inference method described herein. All embodiments and variations of the force inference method may be applied.

[0180] In each implementation, the force map is 2 At least 0.25 force vector per mm 2 At least 0.5 force vector per mm 2 At least 0.75 force vectors per mm 2 At least 1 force vector per mm 2 At least 1.5 force vectors per mm 2 There may be at least two force vectors per

[0181] In each implementation, the force map is 2 Maximum 0.25 force vector per mm 2 Maximum 0.5 force vector per mm 2 Maximum 0.75 force vector per mm 2 Maximum of 1 force vector per mm 2 Maximum of 1.5 force vectors per mm 2 There may be up to two force vectors per

[0182] In one implementation, each force vector comprises a normal force component, a first shear force component, and a second shear force component.

[0183] In one implementation, the first shear force component may correspond to the first shear force and the second shear force component may correspond to the second shear force. The first shear force may be particularly perpendicular to the second shear force. In particular, the shear force components may be perpendicular to each other.

[0184] In one implementation, each force vector has three components in the reference coordinate system.

[0185] In one implementation, a feedforward neural network is trained using the image data plus additional images of the reflective surfaces of the sensor configuration without external impacts.

[0186] In one implementation, a feedforward neural network is trained using images of the wall structure of the sensor configuration as part of the image data.

[0187] In one implementation, a feedforward neural network is trained on position-encoded grayscale gradient images as part of the image data.

[0188] In one implementation, a feedforward neural network is trained on one or more of grayscale gradient images, body images, and reference light patterns.

[0189] In one implementation, the sensor arrangement is a force-sensing sensor arrangement. The sensor arrangement may in particular comprise one or more of the following: an elastically deformable wall, the wall having an outer measurement surface and an inner reflective surface, wherein the reflective surface partially bounds an interior space; a light source configuration including a plurality of light sources and configured to emit light toward an interior space; Image sensor installed inside the interior space is.

[0190] See further the description of the sensor configuration elsewhere in this specification. All of the described embodiments and variations may be applied.

[0191] In one implementation, the sensor arrangement is a force-sensing sensor arrangement. The sensor arrangement may in particular comprise one or more of the following: - a base; a top comprising an elastically deformable wall, the top attached to a base such that the top and base define an interior space, the wall comprising an outer measurement surface and an inner reflective surface, wherein the reflective surface partially bounds the interior space; a light source arrangement including a plurality of light sources attached to a base and configured to direct light toward an interior space; - an image sensor attached to the base within the interior space; is.

[0192] For further possible aspects of the sensor configuration, reference is made to the descriptions elsewhere in this specification. All embodiments and variants are applicable.

[0193] The sensor arrangement may in particular be at the tip of the robot and / or at the manipulating element of the robot, thereby allowing the functionality to be integrated into the robot, although other implementations and applications are also possible.

[0194] The invention further relates to a force inference module for force inference of a force-sensing sensor arrangement, the force inference module being configured to perform the force inference method described herein, for which all implementations and variants are applicable.

[0195] The force reasoning module may be implemented as, for example, a microcontroller, a microprocessor, a field programmable gate array, an application specific integrated circuit, or a computer, and may in particular comprise processing means and storage means, wherein the storage means stores program code that causes the processing means to perform the methods disclosed herein.

[0196] The invention further relates to a sensor arrangement for sensing force, the sensor arrangement comprising one or more of the following: - a base; a top comprising an elastically deformable wall, the top attached to a base such that the top and base define an interior space, the wall comprising an outer measurement surface and an inner reflective surface, wherein the reflective surface partially bounds the interior space; a light source arrangement including a plurality of light sources attached to a base and configured to direct light toward an interior space; an image sensor mounted to a base within the interior space; - the power reasoning module described herein; is.

[0197] With regard to the force inference module, all embodiments and variants described herein may be applied.

[0198] The present invention further relates to a computer program product for performing the methods disclosed herein. The present invention further relates to program code for performing the methods disclosed herein. The present invention further relates to a non-volatile computer-readable storage medium having program code stored thereon, which, when executed by a processor, causes the processor to perform the methods disclosed herein. With regard to the methods, all embodiments and variations disclosed herein may be applied.

[0199] Further aspects and advantages will be apparent to those skilled in the art from the following description of the enclosed drawings, which show: [Brief explanation of the drawings]

[0200] [Figure 1] Figure 1 shows an exploded view of the sensor configuration. [Figure 2] Figure 2 shows a cross-sectional view of the sensor configuration. [Figure 3] Figure 3 shows the top. [Figure 4] Figure 4 shows the body. [Figure 5] Figure 5 shows another structure. [Figure 6] FIG. 6 shows a detection surface with a pattern of light. [Figure 7] FIG. 7 shows a detection surface with another light pattern. [Figure 8] FIG. 8 shows the detection surface with lines of intensity. [Figure 9]FIG. 9 shows a detection surface with lines of different intensities. [Figure 10] FIG. 10 shows multiple light sources with collimators. [Figure 11] Figure 11 shows the body and mold. [Figure 12] FIG. 12 shows a further mould. [Figure 13] Figure 13 shows one of several different types of casts. [Figure 14] Figure 14 shows one of several different types of casts. [Figure 15] Figure 15 shows one of several different types of casts. [Figure 16] Figure 16 shows one of several different types of casts. [Figure 17] Figure 17 shows one of several different types of casts. [Figure 18] Figure 18 shows the cast profile. [Figure 19] Figure 19 shows the generation of a force map. [Figure 20] Figure 20 shows how to create a force map. [Figure 21] Figure 21 shows the force testing setup. [Figure 22] FIG. 22 shows a possible indenter. [Figure 23] Figure 23 shows the force map. [Figure 24] Figure 24 shows the simulated force map. DETAILED DESCRIPTION OF THE INVENTION

[0201] 1 shows a force-sensing sensor arrangement 10 (sensor device). The sensor arrangement 10 comprises a base portion 100 and a top portion 200.

[0202] The base 100 comprises a support structure 110, which may be used, among other things, to mount the sensor arrangement 10 to another entity, such as a robot. The base 100 further comprises a printed circuit board 120, on which the electronic components that control the sensor arrangement 10 are mounted.

[0203] Base 100 further comprises an image sensor 130 located directly above printed circuit board 120. Image sensor 130 is embodied as a color camera capable of detecting light and generating image data in response thereto, where typically the image data generated by image sensor 130 will be further processed by electronic components located on printed circuit board 120.

[0204] The base 100 further includes a wide-angle lens 140 located directly above the image sensor 130. The wide-angle lens 140 is positioned such that all light impinging on the image sensor 130 passes through the wide-angle lens 140. This allows the wide-angle lens 140 to define the field of view of the image sensor 130. The wide-angle lens 140 may be considered part of the image sensor 130.

[0205] The base 100 further comprises a mounting structure 150. The mounting structure 150 is attached directly to the support structure 110 using a number of screws 152 that are used to secure the top 200 to the base 100 such that the top 200 is removable from the base 100. How the connection between the top 200 and the base 100 is made is further described below.

[0206] The base 100 further includes a light source arrangement 160. The light source arrangement 160 includes a support ring 162 attached to the mounting structure 150. The light source arrangement 160 includes a plurality of light sources 164, which are embodied as light emitting diodes in this embodiment. The light sources 164 are positioned to emit light toward an interior space of the top 200, which is further described below with respect to FIG. 2.

[0207] The light sources 164 have different colors, for example red, blue, and green, and how the light emitted from the light sources can be used to detect the force applied to the sensor arrangement 10 is described further below.

[0208] Immediately above the arrangement of light sources 160 is a collimator ring 175. The collimator ring 175 comprises a plurality of collimators 170 embodied as holes protruding vertically through the collimator ring 175. Each collimator 170 is located directly above one light source 164. This allows only light that passes through the collimator 170 to reach the interior space of the top portion 200. The collimators 170 may thus define a plurality of cones of light within the interior space. In particular, each cone of light has a respective cone axis and outer cone angle.

[0209] The top portion 200 comprises an elastically deformable wall 210, which presents a measurement surface 220 on the exterior of the sensor arrangement 10. The measurement surface 220 is a surface on which a force is applied, where the force should be measured, e.g., to provide a force map that depends on the actual applied force. The actual inference of the force is explained further below.

[0210] The top 200 comprises a body 240. The body 240 is surrounded by a wall 210. In Figure 1, the body 240 and the wall 210 are shown separately from each other.

[0211] The body 240 includes a bottom ring 242 and a lattice 244 positioned above the bottom ring 242. The detailed structure of the body 240 will be described further below. The bottom ring 242 includes a plurality of protrusions 246 arranged radially outward and oriented perpendicular to the base 100. These protrusions 246 include respective threaded holes for a plurality of vertically extending screws 152. The screws 152 can be applied from below through these threaded holes. Here, the screws 152 are fixed to respective holes in the mounting structure 150 of the base 100. This allows the top 200 to be removably mounted on the base 100.

[0212] 2 shows a cross-sectional view of the sensor configuration 10 in an installed state. The body 240 is surrounded by the wall 210 but is visible inside the wall 210. This is for illustrative purposes only. In reality, the body 240 surrounded by the wall 210 would not be visible, or at least not be clearly visible.

[0213] 2 shows all components of sensor configuration 10 in their final positions, with the exception of collimator ring 175, which is not shown in FIG. 2 for clarity of viewing source configuration 160.

[0214] 2, the top portion 200 and the base portion 100 define an interior space 12. The interior space 12 is surrounded by a wall 210. The wall 210 defines an inner reflective surface 230 that surrounds the interior space 12.

[0215] As shown in FIG. 2, the light source 164 is disposed in the base 100 so as to emit light toward the interior space 12 .

[0216] Light emitted from light source 164 first passes through collimator 170, which is not shown in Figure 2. Collimator 170 defines the further propagation of the light (specifically, the individual propagation directions and outer cone angles of the cones of light). The light then passes through interior space 12 and reaches reflective surface 230. Reflective surface 230 is diffusely reflective, so that incident light is reflected in all directions with a certain angle-dependent intensity, and the reflection is not specular.

[0217] As shown in FIG. 2 , reflective surface 230 is separated from measurement surface 220 only by wall 210. Wall 210 is made of an elastic material that relays deformation from measurement surface 220 to reflective surface 230. This means that any force applied to measurement surface 220 not only deforms wall 210 at the outer measurement surface 220, but also at the inner reflective surface 230. This deformation of reflective surface 230 locally distorts the reflection of the light beam. As such, measurements of the light beam inside sensor configuration 10 can be used to infer the force applied to measurement surface 220.

[0218] Image sensor 130 is also located within sensor arrangement 10. Image sensor 130 is surrounded by light source 164 such that light emanating from light source 164 and reflected from reflective surface 230 propagates to image sensor 130. On the sensing surface of image sensor 130, the light generates a light pattern indicative of the applied force. Such light pattern has been found to be indicative of the position, amplitude, and direction of the force, as well as the shape and size of the applied indenter. Multiple forces can also be assessed.

[0219] Grating 244 comprises a main portion 248 and a fingernail region 249. Fingernail region 249 retains a thinned portion, which will be further described below. Main portion 248 provides increased stability to wall 210, allowing wall 210 to withstand greater forces and not deform substantially under the influence of external forces and gravity. However, despite body 240, its ability to slightly deform and relay deformations from measurement surface 220 to reflecting surface 230 remains effective.

[0220] FIG. 3 shows the wall 210 separately. The wall 210 has a conical shape as shown. The wall 210 has a rounded shape at the tip. It should be noted that this shape is one example that has proven suitable for multiple applications. However, it is not the only shape possible. Rather, any suitable shape can be used. A measurement surface 220 defines the exterior of the wall 210. If a force is applied to the measurement surface 220, the wall 210 will deform.

[0221] 4 shows the body 240 separated from the rest. For the components already described, reference is made to the description of FIG. 1. In particular, it can be seen that the lattice 244 is formed from a number of wires with relatively large spaces between them. When these spaces are filled with the material of the walls 210, the walls 210 surround the body 240. In this way, the body 240 provides the walls 210 with adequate stability.

[0222] Figure 5 shows a wall 210 according to a second embodiment. In contrast to the embodiment shown in Figure 3, the wall 210 shown in Figure 5 has a thinned area 250. The thinned area 250 has a reduced thickness compared to the rest of the wall 210. The thinned area 250 is surrounded by an edge 255.

[0223] 4 and 5 together, it becomes clear that the fingernail section 249 of the body 240 supports the thin section 250 of the wall 210. The support is generally along the edge 255. This provides greater stability to the thin section 250. Note, however, that the same body design can be used for the top 200 without the thin section 250.

[0224] Thinned areas 250 provide a local increase in sensitivity, particularly with respect to force detection: for example, a force applied to thinned areas 250 leads to a larger deformation of wall 210 and therefore to a larger deformation of reflective surface 230. Thus, a force applied to thinned areas 250 also leads to a larger change in the light pattern detected by image sensor 130.

[0225] 6 shows a pattern of light 132 on a detection surface 131 of an image sensor 130. The detection surface 131 typically comprises a number of pixels, which are not shown in FIG.

[0226] As shown in FIG. 6 , the exemplary light pattern 132 includes eight light spots, designated by reference numerals 133, 134, and 135. The light spots 133, 134, and 135 are at least approximately elliptical in shape. Three of the light spots 133 result from a light source having a first color, e.g., blue. Three of the light spots 134 result from a light source having a second color, e.g., red. Two of the light spots 135 result from a light source having a third color, e.g., green. Note that the number of eight light spots 133, 134, and 135 having three different colors is for illustrative purposes only, and any other number of light spots and colors may be used. In particular, each light spot 133, 134, and 135 may correspond to one light source 164, with light from the light source 164 propagating in a cone through the interior space 12, reflecting off the reflective surface 230, and propagating to the detection surface 131.

[0227] 6 shows a typical light pattern 132 in an undeformed state, i.e., such a light pattern 132 may be seen at the detection surface 131 if no force is applied to the measurement surface 220. Any force inference means, which may be, for example, a neural network, can be trained to detect the absence of force when the light pattern 132 is detected by the image sensor 130 of FIG.

[0228] FIG. 7 shows a further light pattern 132 on the detection surface 131. The light pattern 132 in FIG. 7 corresponds to the application of a force to the measurement surface 220. As can be seen, the two light spots 133, 135 have a deformation 136 with a local color change due to different reflective properties within the reflective surface 230 in response to the applied force. The situation in FIG. 7 can trigger the inference that a force has been applied. In particular, the change in the light pattern 132 is characteristic not only of the force strength, but also of its position, direction, and the shape and size of the indenter applying the force. This is also true for two or more forces that may be applied simultaneously.

[0229] Figure 8 shows a schematic representation of intensity lines 138 of the intensity pattern 137 on the detection surface 131. Each of the intensity lines 138 corresponds to a line of constant light intensity on the detection surface 131. Figure 8 shows the undeformed state, which corresponds to the state shown in Figure 6. Figure 9 shows the deformed state, with the intensity lines 138 having different intensities due to locally different reflective properties of the reflective surface 230. This is also characteristic of the applied force and can therefore be used to infer the force.

[0230] FIG. 10 shows purely diagrammatically an image sensor 130 with a detection surface 131, which is positioned between a light source arrangement 160 with a support ring 162 and a light source 164 with a collimator 170 and collimator ring 175.

[0231] As shown, light emitted from the light source 164 passes through the collimator 170 and propagates further into the interior space 12 in each cone 166. These cones 166 are defined by a central propagation direction 167, which can also be considered the cone axis, and an outer cone angle 168, which defines the maximum extent of the light horizontally relative to the central propagation direction 167. As shown, the collimator 170 is positioned slightly outward relative to the point where the light exits the light source 164, so that the central propagation direction 167 is pointing slightly outward rather than vertically. This ensures that oversaturation at the tip of the apex 200 is prevented. This implementation results in a defined light structure within the interior space 12. Reflected light from a reflective surface 230, not shown in FIG. 8, may propagate to the detection surface 131, where it may be detected for force inference.

[0232] Figure 11 shows a mold 600 for fabricating a top portion for sensor configuration 10. Mold 600 includes a mold body 620 with an opening 610 formed therein. Figure 12 shows a further mold 605, which also has a mold body 620 with an opening 610 formed therein. When mold 600 and further mold 605 are combined, only one opening 610 remains. Opening 610 can be used for fabricating top portion 200 by overmolding body 240.

[0233] Figure 11 shows one stage in the manufacturing process of top portion 200. Opening 610 is filled with body 240. A further mold 605, shown in Figure 12, will be used to form one single opening 610. Body 240 is overmolded with wall material to form wall 210 surrounding body 240.

[0234] The exterior of the formed wall 210 is defined by the opening 610, which will become the measurement surface 220. In other words, the measurement surface 220 conforms to the shape of the opening 610.

[0235] Cast 700 is used to define the interior shape of what will become reflective surface 230. Possible embodiments of cast 700 are shown in Figures 13-17.

[0236] Figure 13 shows a cast 700 according to the first embodiment. The cast 700 includes a support ring 720. From the support ring 720, a main portion 710 has a rounded cross section that tapers toward its tip. In the embodiment of Figure 13, the main portion 710 has a flat outer surface, so that when the cast 700 is used inside the body 240 shown in Figure 11 to define the inner reflective surface 230 of the wall 210, a flat reflective surface 230 is formed.

[0237] FIG. 14 shows a cast 700 according to a second embodiment. In contrast to the embodiment shown in FIG. 13, the cast 700 of FIG. 14 has a plurality of grooves 716 that can be used to provide a specific complementary structure on the reflective surface 230. The grooves 716 form complementary protrusions in the reflective surface 230, which can more significantly increase the change in the light pattern 132. As shown in FIG. 14, the grooves 716 are located near the tip of the main portion 710. For example, the grooves 716 can be applied to the thin area 250 of the wall 210 as shown in FIG. 5.

[0238] Figure 15 shows a cast 700 according to a third embodiment. In the embodiment of Figure 15, the cast 700 has an outer lattice structure 712 that interfaces with a complementary structure on the reflective surface 230, which can improve force detection in many situations.

[0239] FIG. 16 shows a cast 700 according to a fourth embodiment, which is embodied similarly to the embodiment of FIG. 15, but which has a finer lattice structure 712.

[0240] Fig. 17 shows a cast 700 according to a sixth embodiment. In addition to the outer lattice structure 712 or that shown in Fig. 16, the embodiment shown in Fig. 17 has a flat portion 714 without a lattice structure. The flat portion 714 is also located near the tip, which leads to a locally flat reflective surface 230. Such a locally flat reflective surface 230 can be applied, for example, in the thin area 250.

[0241] FIG. 18 shows a typical cross-sectional shape (outline, profile) of a cast 700, such as that shown in FIG. 16. The cross-sectional shape comprises an outer lattice structure 712 with protrusions 713. Between the protrusions 713 are flats that connect to flats on the reflective surface 230. In the illustrated embodiment, the protrusions 713 have a semicircular profile. For example, the distance between the protrusions 713 can be at least 0.1 mm, at least 0.5 mm, at least 1 mm, or at least 2 mm. The distance between the protrusions 713 can be at most 0.5 mm, at most 1 mm, at most 2 mm, or at most 5 mm. Thus, it can range from sub-millimeters to several millimeters. The same value or range can apply to the radius of the protrusions 713. However, other values ​​can also be used.

[0242] Figure 19 shows a schematic diagram of force inference.

[0243] As already mentioned above, the image sensor 130 comprises a detection surface 131. The detection surface 131 has a number of pixels P, each of which is denoted P1, P2, ... Px. Each pixel P is individually capable of colorimetrically detecting light incident on it, thereby making it possible to detect a light pattern 132 on the image sensor 130.

[0244] The output data of the image sensor 130 is provided to a feedforward neural network FFNN, which is an artificial neural network that maps the image data from the image sensor 130 to a force map FM. The force map FM comprises a number of force vectors F1, F2, ...Fx. The force vectors F of the force map FM are described further below.

[0245] The feedforward propagation neural network FFNN may be trained by a training method T. Suitable training is further described below with reference to FIG.

[0246] In principle, a feedforward neural network (FFNN) can detect forces from image data received from the image sensor 130. This can be enhanced by machine learning techniques. A properly trained feedforward neural network (FFNN) has been shown to be able to infer the position, amplitude, and direction of forces, as well as the shape and size of the indenter. Such inference is possible even in the case of the simultaneous application of multiple forces.

[0247] For enhanced force inference, further image data comprising at least one of the following types of observations of undeformed illuminated reflective surfaces, gradient images in grayscale, and images of the body should in each case be input to the feedforward propagation neural network FFNN for force inference and / or training steps.

[0248] FIG. 20 shows a method T for training a feedforward propagation neural network FFNN.

[0249] In a first step T_1, a number of force tests are performed using the sensor arrangement 10, as further described with reference to Figure 21. During each force test, the applied force is measured, and the position of the force on the measurement surface 220 is measured. Additionally, image data corresponding to the light pattern 132 is read out from the image sensor 130.

[0250] In a second step T_2, multiple simulation tests are performed, each corresponding to one force test. Each simulation test is performed using a model of the sensor arrangement 10 that simulates the behavior of the components, in particular the wall 210, when a force is applied. A map FM' of simulated forces is thus calculated for each simulation test.

[0251] In a third step T_3, a feedforward neural network FFNN is trained with the image data of the force trials and the simulated force map FM' calculated for the corresponding simulated trials. In particular, each training sub-step may involve training the feedforward neural network with the light patterns 132 and the corresponding simulated force map FM'. This allows the feedforward neural network FFNN to learn how to map the light patterns 132 to the force map FM.

[0252] It should be noted that the simulated force map FM′ exists only in a simulated test based on a model of the sensor configuration 10. The force map FM exists on the actual measurement surface 220.

[0253] 21 shows a force testing arrangement 500 for performing force tests. The force testing arrangement 500 comprises a base 510. A first arm 520 is disposed on the base 510 and is connected at a joint 530. A second arm 540 is positioned at the joint 530. An electric drive means (not shown) may cause the first arm 520 to rotate on the base 510 and the second arm 540 to swing around the joint 530.

[0254] The second arm 540 has attached thereto the sensor arrangement 10 described above, which may be rotated about the axis of the second arm 540. The sensor arrangement 10 may thus be positioned by the force testing arrangement 500.

[0255] There is also a top portion 550, on which a force sensor 560 is attached. The force sensor 560 is connected with an indenter 800. The indenter 800 remains in a substantially unchanged position. The force testing arrangement 500 allows the sensor arrangement 10, and in particular its measurement surface 220, to be brought into contact with the indenter 800 so that a force can be applied. This force can be measured by the force sensor 560.

[0256] In one preferred implementation, arms 520, 540 are used to select an intended location on measurement surface 220 where indenter 800 should contact measurement surface 220. Indenter 800 can then be moved by moving tip 550 in three dimensions, thus applying a force to measurement surface 220 that may have both normal and shear force components. Sensor arrangement 10 may remain in place during the application of the force. However, other implementations of force testing are possible, particularly with regard to component movement. For example, tip 550 may be moved simultaneously with arms 520, 540. Alternatively, only arms 520 and 540 may be used to apply the force.

[0257] The position where the indenter 800 contacts the measurement surface 220 can be calculated using machine variables or kinematic models, but observations can also be made with a camera.

[0258] FIG. 22 shows schematically four different indenter 800 shapes, which may be physical indenters 800 for use in the force testing arrangement 500, or may be simulated indenters 800′ in simulated testing as further described below with respect to FIG. 24.

[0259] FIG. 22a shows an indenter 800 having a flat shape at its contact portion with the measurement surface 220. FIG. 22b shows an indenter 800 having a contact portion in the form of a tip. FIG. 22c shows an indenter 800 having a contact portion in the form of a hemisphere. FIG. 22d shows an indenter 800 having a contact portion of the same shape as the indenter 800 shown in FIG. 1 but with a smaller size. Using such different indenters 800 allows the training of the feedforward propagation neural network FFNN to be optimized for such different shapes, which means that the performance of the feedforward propagation neural network FFNN trained with such different indenters 800 increases with respect to the reconstructed forces applied by indenters 800 having different indenter shapes. In other words, as an example, the force map FM reconstructed after applying a flat-shaped indenter 800 will be different from the force map FM reconstructed after applying a hemispherical indenter 800.

[0260] FIG. 23 shows the sensor configuration 10 with a schematic illustration of a force map FM. The force map FM comprises a number of force vectors F located all around the circumference of the measurement surface 220. Although two force vectors F are shown in FIG. 23, many more force vectors F may be used in a typical implementation. For example, mm 2 A single force vector F per unit area can be used in the exemplary implementation.

[0261] Each force vector F has a normal force component F N , the first shear force component F S1 and the second shear force component F S2 The normal force component F n gives the value of the applied force, i.e., the normal force component perpendicular to the local orientation of the measurement surface 220. The shear force component F S1 , F S2 gives the value of the shear force acting on the measurement surface 220 at each point. The shear forces are typically parallel to the local orientation of the measurement surface 220 and are typically perpendicular to each other and to the normal force. This may particularly relate to the undeformed orientation of the measurement surface 220, which may define the direction of the force vector F, especially its normal component.

[0262] Each force vector F thus gives the magnitude and direction of a force applied to a particular point on the measurement surface 220. Such a force may originate, for example, from an indenter 800, one such indenter 800 being shown by way of example in Figure 23. When a force is applied, the measurement surface 220 deforms slightly.

[0263] Note that other definitions of the force vector F may also be used, for example, only the normal force component may be evaluated, or the shear force may have an alternative definition.

[0264] 24 shows the corresponding case using a simulated force map FM'. The reference symbol is indicated by an apostrophe ('). In the case of a simulated force map FM', the simulated force vectors F' of such a simulated force map FM' on the simulated measurement surface 220' each have a simulated component, e.g., a normal force component F' N , the first shear force component F' S1 and the second shear force component F' S2 Such a map of simulated forces FM' is calculated in particular in a simulation run carried out on the model as described above.

[0265] There is also a simulated indenter 800' shown in Figure 24. The simulated indenter 800' is applied to the simulated test with the same magnitude and direction and at the same position as the real indenter 800 in the corresponding force test. Using the model, a simulated force map FM' is calculated and used to train the feedforward neural network FFNN.

[0266] The steps mentioned in the method of the present invention can be performed in a given order. However, they may be performed in another order, as long as this is technically reasonable. The method of the present invention can be implemented in embodiments, for example, using a specific combination of steps, such that no further steps are performed. However, other steps, including steps not mentioned, may also be performed.

[0267] It should be noted that although several features (of the present invention) may be used or implemented independently of each other, for example, for the sake of clarity, features are described in combination within the claims and the description. Those skilled in the art will recognize that such features can be combined with other features or that there may be combinations of features independent of each other.

[0268] References in the dependent claims may indicate preferred combinations of features respectively, but do not exclude other combinations of features. The present application provides the following aspects, for example: (Point 1) A sensor arrangement (10) for detecting a force, said sensor arrangement (10) comprising a base (100); A top (200) having an elastically deformable wall (210), the top portion (200) is attached to the base portion (100) such that the top portion (200) and the base portion (100) define an interior space (12); the wall (210) having an outer measurement surface (220) and an inner reflective surface (230), wherein the reflective surface (230) partially bounds the interior space (12); a light source arrangement (160) comprising a plurality of light sources (164) attached to the base (100) and configured to emit light toward the reflective surface (230); an image sensor (130) having a detection surface (131) that views at least a portion of the reflective surface (230); A sensor configuration (10) comprising: (Point 2) Each of said light sources (164) has a respective color; The light source arrangement (160) comprises a plurality of light sources (164) having at least two or three different colors. A sensor configuration (10) according to aspect 1. (Point 3) A sensor configuration (10) according to aspect 1 or 2, wherein the light source (164) and the detection surface (131) are arranged such that light emitted by the light source (164) and reflected by the reflective surface (230) generates a light pattern (132) on the detection surface (131). (Point 4) Aspects 4. The sensor arrangement (10) according to any one of aspects 1 to 3, wherein the pattern of light (132) on the detection surface (131) changes in response to a deformation of the measurement surface (220). (Point 5) A sensor arrangement (10) according to any one of aspects 1 to 4, wherein the color distribution of the reflected light from the reflecting surface (230) on the detection surface (131) changes in response to deformation of the measurement surface (220). (Point 6) Aspects 6. The sensor arrangement (10) according to any one of aspects 1 to 5, wherein the reflective surface (230) is diffusely reflective. (Point 7) A sensor configuration (10) according to any one of aspects 1 to 6, wherein the image sensor (130) is mounted on at least one of the base (100) and within the interior space (12). (Point 8) A sensor configuration (10) according to any one of aspects 1 to 7, wherein the light source (164) is arranged surrounding the image sensor (130). (Point 9) A sensor configuration (10) according to any one of aspects 1 to 8, wherein the light source (164) is arranged to emit light so as to produce a distribution of reflected light on the detection surface (131) that does not produce an intensity exceeding the saturation density. (Point 10) A sensor arrangement (10) according to any one of aspects 1 to 9, wherein the light source (164) is a light emitting diode. (Point 11) A sensor configuration (10) according to any one of aspects 1 to 10, wherein the sensor configuration (10) comprises a plurality of collimators (170), each collimator (170) being assigned to one light source (164) and defining at least one of an illumination angle and a cone of emitted light. (Point 12) A sensor arrangement (10) according to aspect 11, wherein one, some or all of the collimators (170) are arranged off-center with respect to the assigned light source. (Point 13) 13. The sensor arrangement (10) according to aspect 11 or 12, wherein the collimator (170) is in the form of a hole in a collimator ring (175). (Point 14) 14. The sensor configuration (10) of any one of aspects 1 to 13, wherein at least one of the light source (164) and the collimator (170) is positioned to cover with light at least 80%, at least 85%, at least 90% or 100% of the reflective surface (230) or the intended measurement area or area within the measurement surface (220) or the intended measurement surface. (Point 15) A sensor configuration (10) according to any one of aspects 1 to 14, wherein at least one of the light sources (164) and the collimator (170) is configured such that at least 60%, at least 70%, at least 80%, at least 90% or 100% of the reflective surface (230) is directly illuminated only by light from at least one of up to four of the light sources (164) and at least two of the light sources (164). (Point 16) 16. The sensor arrangement (10) according to any one of aspects 11 to 15, wherein the collimator (170) has a collimator hole diameter of at least 0.8 mm. (Point 17) 17. The sensor arrangement (10) according to any one of aspects 11 to 16, wherein the collimator (170) has a collimator hole diameter of up to 4 mm. (Point 18) Aspects 18. The sensor arrangement (10) according to any one of aspects 1 to 17, wherein the light sources (164) are arranged in a circle on the base (100). (Point 19) A sensor configuration (10) according to any one of aspects 1 to 18, wherein the detection surface (131) is configured and arranged to observe at least 70%, at least 80%, at least 90% or the entire reflective surface (230). (Point of View 20) The image sensor (130) having a color camera sensor; The image sensor (130) is color sensitive. 20. The sensor arrangement (10) according to any one of aspects 1 to 19, (Point of View 21) A sensor configuration (10) according to any one of aspects 1 to 20, wherein the image sensor (130) comprises a plurality of pixels (P), each pixel (P) being configured to independently detect light. (Point of View 22) 22. The sensor arrangement (10) according to any one of aspects 1 to 21, wherein the detection surface (131) faces towards the reflecting surface (230). (Point of View 23) Aspects 1 to 23. The sensor configuration (10) according to any one of aspects 1 to 22, wherein the detection surface (131) is parallel to an inner surface of the base (100). (Point of View 24) A sensor arrangement (10) according to any one of aspects 1 to 23, wherein the detection surface (131) is configured to detect at least one of a plurality of light patterns (132) and a plurality of images. (Point of View 25) A sensor configuration (10) according to any one of aspects 1 to 24, wherein the image sensor (130) is configured to detect at least one of a plurality of light patterns (132) and a plurality of images having a frame rate. (Point of View 26) 26. The sensor arrangement (10) according to aspect 25, wherein the frame rate is at least 10 fps, at least 20 fps, at least 30 fps, at least 50 fps, or at least 100 fps. (Point of View 27) 27. The sensor configuration (10) according to aspect 25 or 26, wherein the frame rate is up to 30 fps, up to 50 fps, up to 100 fps, or up to 200 fps. (Point of View 28) A sensor configuration (10) according to any one of aspects 1 to 27, wherein the image sensor (130) comprises a wide-angle lens (140) or a fisheye lens optically positioned between the reflective surface (230) and the detection surface (131). (Point of View 29) Aspects 1 to 29. The sensor configuration (10) according to any one of aspects 1 to 28, wherein the internal space (12) is a hollow space. (Point of View 30) A sensor configuration (10) according to any one of aspects 1 to 29, wherein at least one of the light source (164) and the collimator (170) is configured to emit light in each cone (166). (Point of View 31) A sensor configuration (10) as described in aspect 30, wherein one, some or all of the cones (166) have cone axes (167) inclined outward at an angle greater than 0° relative to at least one of a common axis and a normal axis of the base (100). (Point of View 32) A sensor configuration (10) according to aspect 30 or 31, wherein one, some or all of the cones (166) have cone axes (167) inclined outward by up to 10° relative to at least one of a common axis and a normal axis of the base (100). (Point of View 33) A sensor configuration (10) according to any one of aspects 30 to 32, wherein one, some or all of the cones (166) have an outer cone angle (168) of at least 35° and at most 80°. (Point of View 34) A sensor configuration (10) according to any one of aspects 30 to 33, wherein the cone (166) partially overlaps a plane perpendicular to the common axis, the overlap being determined by the distance between the plane and the base. (Point of View 35) Aspects 1 to 35. The sensor arrangement (10) according to any one of aspects 1 to 34, wherein the reflective surface (230) is covered with at least one of a pattern and a plurality of trackable objects. (Point of View 36) Aspects 1 to 35. The sensor arrangement (10) according to any one of aspects 1 to 34, wherein the reflective surface (230) is a smooth surface. (Point of View 37) Aspects 1 to 37. The sensor arrangement (10) of any one of aspects 1 to 36, wherein the sensor arrangement (10) has a common axis and the top (200) is generally circular about the common axis. (Point of View 38) Aspects 1 to 37. The sensor arrangement (10) according to any one of aspects 1 to 36, wherein the sensor arrangement (10) has a common axis and the top (200) is partially circular around the common axis. (Point of View 39) 39. The sensor arrangement (10) according to aspect 37 or 38, wherein the detection surface (131) is perpendicular to the common axis. (Point of View 40) 40. The sensor arrangement (10) according to any one of aspects 37 to 39, wherein the light source (164) is configured to emit light parallel to the common axis. (Point of View 41) Aspects 1 to 41. The sensor configuration (10) of any one of aspects 1 to 40, wherein different portions of the reflective surface (230) are illuminated by the light source (164) from different directions and with different colors. (Point of View 42) A sensor configuration (10) according to any one of aspects 1 to 41, wherein the top portion (200) is tapered with an outer diameter that decreases as the distance from the base portion (100) increases. (Point of View 43) 43. The sensor arrangement (10) according to any one of aspects 1 to 42, wherein the wall (210) is configured to relay deformation from the measurement surface (220) to the reflecting surface (230). (Point of View 44) Aspects 1 to 44. The sensor arrangement (10) according to any one of aspects 1 to 43, wherein the top (200) is removably attached to the base (100). (Point of View 45) A sensor configuration (10) according to any one of aspects 1 to 44, wherein the top portion (200) is removably attached to the base portion (100) by at least one of a bayonet method, a screw connection, and a pair of tangs and corresponding notches. (Point of View 46) 46. ​​The sensor arrangement (10) according to any one of aspects 1 to 45, wherein the top (200) comprises only the wall (210). (Point of View 47) Aspects 1 to 46. The sensor arrangement (10) according to any one of aspects 1 to 45, wherein the top (200) comprises a body (240) arranged inside the wall (210). (Point of View 48) A sensor arrangement (10) according to aspect 47, wherein the body (240) is in the form of a grid. (Point of View 49) 49. The sensor arrangement (10) according to aspect 47 or 48, wherein the body (240) is made of steel, stainless steel or aluminum. (Point of View 50) 50. The sensor arrangement (10) according to any one of aspects 47 to 49, wherein the body (240) is rigid or semi-rigid. (Point of View 51) Aspects 1 to 50. The sensor arrangement (10) according to any one of aspects 1 to 50, wherein the wall (210) comprises a wall material. (Point of View 52) A sensor arrangement (10) according to aspect 51, wherein the wall material comprises an elastomer. (Point of View 53) 53. The sensor configuration (10) of aspect 51 or 52, wherein the wall material comprises at least one of aluminum powder and aluminum flakes. (Point of View 54) The sensor arrangement (10) according to aspects 47 and 51 or according to one of the aspects dependent thereon, wherein the body (240) is surrounded by the wall material. (Point of View 55) A sensor arrangement (10) according to aspects 47 and 51 or according to one of the aspects dependent thereon, wherein the body (240) is completely overmolded with the wall material. (Point of View 56) A sensor arrangement (10) according to aspects 47 and 51 or according to one of the aspects dependent on aspects 47 and 51, wherein the body (240) is partially overmolded with the wall material. (Point of View 57) the wall (210) has a thickness of at least 0.8 mm or at least 1.2 mm; The wall (210) has a thickness of at most 4 mm or at most 5 mm. 57. The sensor configuration (10) according to any one of aspects 1 to 56, wherein the sensor configuration (10) is at least one of the following: (Point of View 58) A sensor configuration (10) according to any one of aspects 1 to 57, wherein the wall (210) has a thinned area (250) that is thinner than areas of the wall (210) other than the thinned area (250). (Point of View 59) A sensor configuration (10) according to aspect 58, wherein the thin area (250) is disposed opposite the base (100). (Point of View 60) 60. The sensor arrangement (10) according to aspect 58 or 59, wherein the thinned area (250) is shaped like a fingernail. (Point of View 61) 61. The sensor configuration (10) according to any one of aspects 58 to 60, wherein the thin area (250) is less than a quarter of the measurement surface (220). (Point of View 62) The thinned area (250) A maximum thickness of 0.8 mm or a maximum thickness of 1.2 mm; having a thickness of at most 30% or at most 50% of the thickness of the wall (210) outside the thinned area (250); 62. The sensor configuration (10) according to any one of aspects 58 to 61, wherein at least one of (Point of View 63) 63. The sensor arrangement (10) of any one of aspects 1 to 62, wherein the sensor arrangement (10) is at least one of a robotic fingertip and a robotic operating element. (Point of View 64) 1. A method of making a top portion (200) for a sensor arrangement (10), said method comprising: providing a body (240) made of a body material, said body (240) enclosing an interior space (12); covering the body (240) with a wall material such that the wall material creates an elastically deformable wall (210), the wall (210) defining an outer measurement surface (220) and an inner reflective surface (230), the reflective surface (230) defining the interior space (12); A method of making a top portion (200) for a sensor arrangement (10), comprising: (Point of View 65) 65. The method of claim 64, wherein the body material is stronger than the wall material. (Point of View 66) 66. The method of claim 64 or 65, wherein the body (240) is more rigid than the wall material. (Point of View 67) Aspect 67. The method of any one of aspects 64 to 66, wherein providing the body (240) comprises three-dimensionally shaping the body (240). (Point of View 68) 68. The method of any one of aspects 64 to 67, wherein the body (240) is made of a plurality of wires, and a plurality of openings are formed between the plurality of wires. (Point of View 69) 69. The method of any one of aspects 64 to 68, wherein the body (240) is dome-shaped. (Point of View 70) 70. The method of any one of aspects 64 to 69, wherein the covering is made by overmolding. (Point of View 71) 71. The method of any one of aspects 64 to 70, wherein inside the top portion (200), the interior space (12) is a hollow space. (Point of View 72) 72. The method of any one of aspects 64 to 71, further comprising covering the reflective surface (230) with at least one of a pattern and a plurality of trackable objects. (Point of View 73) 73. The method of aspect 72, wherein a structured cast (700) is applied to create the pattern. (Point of View 74) The body (240) has a common axis, 74. The method of any one of points 64 to 73, wherein the body (240) is partially or wholly circular about the common axis. (Point of View 75) 75. The method of any one of aspects 64 to 74, wherein the body (240) is tapered with an outer diameter that decreases toward a tip. (Point of View 76) 76. The method of any one of aspects 64 to 75, wherein the wall material is configured to relay deformation from the measurement surface (220) to the reflecting surface (230). (Point of View 77) 77. The method of any one of aspects 64 to 76, wherein the body (240) is in the form of a lattice. (Point of View 78) 78. The method of any one of aspects 64 to 77, wherein the body (240) is made from steel, stainless steel, or aluminum. (Point of View 79) 79. The method of any one of aspects 64 to 78, wherein the body (240) is rigid or semi-rigid. (Point of View 80) 80. The method of any one of aspects 64 to 79, wherein the wall material comprises an elastomer. (Point of View 81) 81. The method of any one of aspects 64 to 80, wherein the wall material comprises at least one of aluminum powder and aluminum flake. (Point of View 82) 82. The method of any one of aspects 64 to 81, wherein the body (240) is covered with the wall material such that the wall material surrounds the body (240). (Point of View 83) the wall material having a thickness of at least 0.8 mm or at least 1.2 mm; the wall material has a thickness of up to 4 mm or up to 5 mm; 83. The method of any one of aspects 64 to 82, wherein the wall material is formed so as to be at least one of: (Point of View 84) 84. The method of any one of points 64 to 83, wherein the wall material is formed to have a thinned area (250) that is thinner than the wall material outside the thinned area (250). (Point of View 85) 85. The method of claim 84, wherein the thinned area (250) is located near the tip of the body (240). (Point of View 86) 86. The method of claim 84 or 85, wherein the thinned area (250) is shaped to conform to a fingernail. (Point of View 87) 87. The method of any one of aspects 84 to 86, wherein the thin area (250) is less than a quarter of the measurement surface (220). (Point of View 88) The thinned area (250) A maximum thickness of 0.8 mm or a maximum thickness of 1.2 mm; having a thickness of at most 30% or at most 50% of the thickness of the wall (210) outside the thinned area (250); 88. The method of any one of aspects 84 to 87, wherein at least one of (Point of View 89) 89. The method of any one of aspects 64 to 88, wherein the body (240) is covered with the wall material so that the wall material is removably attached to the body (240). (Point of View 90) A method of making a sensor arrangement (10), said method comprising: providing a base (100); creating a top portion (200) having an interior space (12); attaching a light source arrangement (160) comprising a plurality of light sources (164) to the base (100); attaching an image sensor (130) to the base (100); covering the base (100) with the top (200) so that the light source (164) is positioned to emit light toward the interior space (12) and the image sensor (130) is located within the interior space (12); A method of making a sensor arrangement (10), comprising: (Point of View 91) 91. The method according to aspect 90, wherein the top (200) is made according to any one of aspects 64 to 89. (Point of View 92) The light source (164) is mounted so that the light source (164) surrounds the image sensor (130). 92. The method according to aspect 90 or 91. (Point of View 93) A method according to any one of aspects 90 to 92, further comprising the step of disposing a plurality of collimators (170) on the light source (164) to define at least one of illumination angles of the emitted light and a plurality of cones of the emitted light. (Point of View 94) Aspect 94. The method of any one of aspects 90 to 93, wherein the sensor arrangement (10) produced is according to any one of aspects 1 to 63. (Point of View 95) A sensor configuration (10) manufactured according to any one of aspects 1 to 63 or according to the method described in any one of aspects 90 to 94, further comprising an electronic control module configured to implement the force inference method of the sensor configuration (10). (Point of View 96) A sensor configuration (10) as described in aspect 95, wherein the control module is configured to implement the method of force inference to provide a force map (FM) of the measurement surface (220), the force map (FM) comprising a plurality of force vectors (F). (Point of View 97) The force map (FM) mm 2 At least 0.25 force vector per mm 2 At least 0.5 force vector per mm 2 At least 0.75 force vectors per mm 2 At least 1 force vector per mm 2 At least 1.5 force vectors per mm 2 and providing at least two force vectors per mm 2 Maximum 0.25 force vector per mm 2 Maximum 0.5 force vector per mm 2 Maximum 0.75 force vector per mm 2 Maximum 1 force vector per mm 2 Maximum 1.5 force vectors per mm 2 With up to two force vectors per The sensor configuration (10) according to aspect 96, wherein the sensor configuration (10) is at least one of: (Point of View 98) Each force vector (F) is divided into normal force components (F N ) and the first shear force component (F S1 ) and the second shear force component (F S2 98. The sensor arrangement (10) according to aspect 95 or 97, comprising: (Point of View 99) The first shear force component (F S1 ) corresponds to the first shear force, and the second shear force component (F S2 ) corresponds to the second shear force, The sensor configuration (10) of aspect 98, wherein the first shear force is perpendicular to the second shear force. [Explanation of symbols]

[0269] 10 Sensor Configuration 12 Interior Space 100 base 110 Support structure 120 Printed Circuit Board 130 Image Sensor 131 Detection Surface 132 Light Pattern 133 Spot of Light 134 Spot of Light 135 Spot of Light 136 Deformed part 137 Light Intensity Patterns 138 Lines of Strength 139 Deformation Zone 140 wide-angle lens 150 Mounting structure 152 screws 160 Light Source Configuration 162 Support ring 164 light source 166 Cone 167 Central Propagation Direction 168 outer cone angle 170 Collimator 175 Collimator ring 200 Top 210 Wall 220 Measurement Surface 230 Reflective Surface 240 skeleton 242 Bottom ring 244 Lattice 246 Protrusions (multiple) 248 Main part 249 Fingernail area 250 Thin Area 255 En 500 Force Test Configuration 510 base 520 First Arm 530 joints 540 Second Arm 550 Top 560 Force Sensor 600 mold 605 More Molds 610 Aperture 620 mold body 700 (Cast) Cast 710 Main part 712 Lattice structure 713 protrusions (multiple) 714 Flat part 716 Grooves (plural) 720 Support ring 800 indenter P pixel FFNN Feedforward Neural Network FM Force Map T Training method F force vector F N Normal force component F S Shear force components Apostrophe "'" Mock test elements

Claims

1. A force sensing sensor arrangement (10), comprising: a base (100); A top (200) comprising an elastically deformable wall (210), the top portion (200) is attached to the base portion (100) such that the top portion (200) and the base portion (100) define an interior space (12); the wall (210) having an outer measurement surface (220) and an inner reflective surface (230), wherein the reflective surface (230) partially bounds the interior space (12); a light source arrangement (160) comprising a plurality of light sources (164) attached to the base (100) and configured to emit light toward the reflective surface (230); an image sensor (130) having a detection surface (131) that observes at least a portion of the reflective surface (230); A sensor arrangement (10) comprising:

2. Each of said light sources (164) has a respective color; The light source arrangement (160) comprises a plurality of light sources (164) having at least two or three different colors. The sensor arrangement (10) according to claim 1.

3. 3. The sensor arrangement (10) of claim 1 or 2, wherein the light source (164) and the detection surface (131) are arranged such that light emitted by the light source (164) and reflected by the reflective surface (230) generates a light pattern (132) on the detection surface (131).

4. a pattern of light (132) on the detection surface (131); a color distribution of the reflected light from the reflecting surface (230) on the detecting surface (131); The sensor arrangement (10) of any one of claims 1 to 3, wherein at least one of the following changes in response to a deformation of the measurement surface (220).

5. The sensor arrangement (10) of any one of claims 1 to 4, wherein the reflective surface (230) is diffusely reflective.

6. The sensor arrangement (10) of any one of claims 1 to 5, wherein the image sensor (130) is mounted on the base (100) and / or within the interior space (12).

7. The sensor arrangement (10) of any one of claims 1 to 6, wherein the light source (164) is arranged surrounding the image sensor (130).

8. 8. The sensor arrangement (10) of claim 1, wherein the sensor arrangement (10) comprises a plurality of collimators (170), each collimator (170) being assigned to one light source (164) and defining at least one of an illumination angle and a cone of emitted light.

9. 9. The sensor arrangement (10) of claim 8, wherein one, some or all of the collimators (170) are positioned off-center with respect to the assigned light source.

10. The sensor arrangement (10) according to claim 8 or 9, wherein the collimator (170) is in the form of a hole in a collimator ring (175).

11. The sensor arrangement (10) according to any one of claims 1 to 10, wherein the interior space (12) is a hollow space.

12. 11. The sensor arrangement (10) of any one of claims 8 to 10, wherein at least one of the light source (164) and the collimator (170) is configured to emit light into each cone (166).

13. one, some or all of the cones (166) have cone axes (167) inclined outward at an angle greater than 0° relative to at least one of a common axis and a normal axis of the base (100); or one, some or all of the cones (166) have cone axes (167) inclined outwardly by up to 10° relative to at least one of the common axis and the normal axis of the base (100); A sensor arrangement (10) according to claim 12.

14. 14. The sensor arrangement (10) of any one of claims 1 to 13, wherein the reflective surface (230) is covered with at least one of a pattern and a plurality of objects trackable by the image sensor.

15. The sensor arrangement (10) of any one of claims 1 to 13, wherein the reflective surface (230) is a smooth surface.

16. The sensor arrangement (10) of any one of claims 1 to 15, wherein the wall (210) is configured to relay deformations from the measurement surface (220) to the reflecting surface (230).

17. The sensor arrangement (10) of any one of claims 1 to 16, wherein the top portion (200) is removably attached to the base portion (100).

18. The sensor arrangement (10) of any one of claims 1 to 17, wherein the top (200) comprises a body (240) arranged inside the wall (210).

19. 19. The sensor arrangement (10) of any one of claims 1 to 18, wherein the wall (210) comprises a thinned area (250) that is thinner than the wall (210) outside the thinned area (250).

20. A method of making a top (200) for a sensor arrangement (10), said method comprising: providing a body (240) made of a body material, said body (240) enclosing an interior space (12); covering the body (240) with a wall material such that the wall material creates an elastically deformable wall (210), the wall (210) defining an outer measurement surface (220) and an inner reflective surface (230), the reflective surface (230) defining the interior space (12); A method of making a top portion (200) for a sensor arrangement (10), comprising:

21. A method of making a sensor arrangement (10), said method comprising: Providing a base (100); Creating a top (200) with an interior space (12); Attaching a light source arrangement (160) comprising a plurality of light sources (164) to the base (100); Attaching an image sensor (130) to the base (100); covering the base (100) with the top (200) so that the light source (164) is positioned to emit light toward the interior space (12) and the image sensor (130) is located within the interior space (12); A method of making a sensor arrangement (10) comprising:

22. The top (200) is subjected to the following steps: providing a body made of a body material, the body enclosing an interior space; encasing the body with a wall material, the wall material creating an elastically deformable wall defining an exterior measurement surface and an interior reflective surface, the interior reflective surface delimiting the interior space; 22. The method of claim 21 , wherein the polymer is prepared according to the following formula:

Images