Patent Application
20200246965
The preferred field of application of the present invention concerns the control of the ambulation of the self-moving platforms, with particular reference to the development of robotized systems suitable to ambulate in environments which are not necessarily known.
One of the most important aspects in the control of the ambulation of a robotized platform (or briefly, “a robot”) consists of the management of the possible bumps or contacts that can affect this robot when it moves, especially in an environment not completely modeled in the memory the robot itself. As it is in the case in which a moving robotized platform is activated and it can move freely, and therefore not on a predetermined path (such as a rail or a guide), within an environment that can also present obstacles variable in time: an environment as it can be a room of a house or an office.
A first technological approach, which enables the management of such situations, provides for the self-moving robotized platform is equipped with sensors able, somehow, “to perceive” (or “to see”) the environment in which it is moving. In this case, through the processing of the information received by these sensors, it is possible to prevent collisions with objects present in the environment.
The most important contraindication associated to this technology relates to the complexity of the processing, as they must substantially perform image recognition of scenes of a certain complexity, and such complexity is the greater the more the environments are not known a priori.
Both some operative limitations of this approach and some complexity factors depend on the technology of the sensors. For example, the infrared sensors do not properly perceive the presence of black, transparent or reflective obstacles, while the ultrasonic sensors incorrectly perceive very thin obstacles, or having surfaces not perpendicular to the ultrasound wave. In the case of use of vision sensors, instead, the complexity is in the estimation of distances, which is substantially impossible with monoscopic signals, and it is excessively complex with stereoscopic signals.
Then, the sensors necessary to the perception of a scene typically involves significant costs and energy consumption, and also require a considerable design complexity.
The complexity of required processing is further increased in the case of management of any contact that is not determined by the motion of the robot: for example, the case in which the robot is bumped or pushed, deliberately or accidentally, by another entity also in movement.
In general, it can be concluded that, to properly handle all the possible cases, the perception of sensors should ensure an omnidirectional coverage and provide a huge amount of information; and the processing required to synthesize a correct and reliable interpretation of events would end up to be, therefore, more and more complicated.
A second technological approach is based on sensors of contact (or pressure), which, given their simplicity, can be realistically disposed over the entire surface of the robotized platform, or over a large part of it.
The sensors of contact, obviously, can manage collisions only after they have occurred. Therefore, it is important to point out that the two technological approaches can be considered, in a certain sense, complementary, and a self-moving robotized platform can, in principle and when necessary, be controlled by means of both the above mentioned technologies, which are not necessarily alternatives, and can coexist in the same platform.
A further interesting application, made possible by the use of sensors of contact, which goes beyond to the management of collisions, is the ability to manually manage the movements of a self-moving robotized platform. In fact, by means of the sensors of contact it is possible to detect when the platform is pushed, for example, with one hand, and therefore it is possible to activate a motorized movement of the robotized platform in the direction consistent with the thrust received.
The manual control of self-moving platforms, in particular for some applications, has some advantages in terms of performance, compared with other control methodologies. For further information about the advantages of the manual controls of powered robotized platforms, please refer to: Jevtic A., Doisy g., Parmet Y., And Edan Y., “Comparison of interaction Modalities for Mobile Robot Guidance: Direct Physical Interaction, Person Following, and Direct Pointing”—IEEE Transactions on Human-Machine Systems—2168-2291 2015 IEEE.
The development of self-moving robotized platforms, capable of moving on generic, non-pre-defined and non-infrastructured paths, at the state of the art, is still a research issue, which is susceptible, and in need, of further evolutionary steps, before becoming a technology mature enough and suitable to be offered commercially, at least on a medium scale.
The present invention is inserted in this development context, it focuses in the solution of problems related to collisions and, more in general, to the contacts that a self-moving robotized platform may have with elements of the external environment.
Typically, the self-moving robotized platforms move on wheels, which may be of various types, may have more axes of rotation permitted, and may assume different forms (from classical disk shape to spherical forms, with different systems of constraint and/or transmission of the movement). Therefore, an initial approach which detects potential collisions can be based on the analysis of the resistant couples observed on the wheel. In fact, if a platform collides with an obstacle, the drive torque that before the impact was sufficient to move the platform, becomes insufficient, or the motion undergoes an appreciable slowdown. Even any manual thrusts, eventually exerted on the robotized platform, would be identifiable to some extent, since they would appear as torque variations detectable on the wheels.
Then, it is possible to address the problem of the management of the motion of a self-moving robotized platform through the simple use of information obtainable from the analysis of the torques observable on its wheels.
The advantages of this approach are evident since it does not require the use of special sensors: in fact, the control of the movement must necessarily take place through the control of the driving torques on the wheels, and therefore the management of collisions and contacts can take place through a calculation based on the continuous analysis of information that are necessarily available, at least in part, for the control of the same platform, as the torques on the wheels and their angular speed.
This approach is not deepened because, despite being advantageously applicable in a certain variety of situations, a general use has significant drawbacks related to the fact that it must be based on an accurate modeling of the inertial system of the same platform and of the land on which the platform moves. In fact, a variation of driving torque can be determined from a weight change (for example, the platform can carry variable loads), or from a change in form (for example the platform may have the extendable arms which determine balancing variations), or the land may have different friction or small inclines not always adequately modeled, and in all these latter cases are possible interpretive misunderstandings, even coarse.
For completeness, it matters to note that it is also possible to conceive methods that do not adopt a particularly accurate modeling. They are such cases in which it is not requested a particular maneuvering accuracy: methods in which it is possible to accept a compromise that favors simplicity of implementation, or robustness, with respect to precision.
An example, at the state of the art, which is based on the analysis of the torques on the wheels of a self-moving platform to handle its ambulation, is given by the “Trikey robots” project conducted at the University of Texas in the United States of America.
Obviously, thrusts exerted on a platform, or eventual accidental collisions, can be detected by the insertion of sensors of force or of moment applied on different parts of the platform's structure, and not necessarily (or only) on the wheels. The case of sensors of torque applied on the wheels is particularly interesting (and it is known in many applications) since the measurement of torques on the wheels, in practice, is automatically available for the fact that the control of the platform's movement is exerted on the wheels. The use of sensors placed elsewhere in the structure of the platform, however, gives rise to alternative variants; then, in these variants the estimation of a pressure exerted on the platform is processed from measurements provided by some sensors localized in some specific parts of the platform.
In all such cases, and without going into the details of this approach, it matters to indicate how this approach is rarely suitable to detect low-intensity collisions, especially when the mobile platform is relatively heavy.
In order to manage contacts and collisions that can affect a self-moving robotized platform, more in general and independently from the inertial modeling, i.e. a modeling that is very difficult to be formalized, complex to be used and harbinger of ambiguous interpretations, it is appropriate to use robotized platforms covered with sensors sensitive to the touch.
There exist quite mature and reliable technologies, which allow to realize a thin cover, which is sensitive to a touch. In the context of technologies for robotics, such coverages are commonly referred to with the term “skin”, precisely to indicate a thin cover that return a signal as a function of the contact. Ideally, in fact, it can be assumed that the skin is the seat of the sense of touch: for which, in the context of robotics, the term “skin” usually indicates a tactile sensor system.
As said, there are different technologies with which a “skin” can be created, and they can be roughly divided into two families: Boolean “skins”, and analog “skins”. The Boolean “skins” return a two values signal, representative of the fact that there is a contact or not. These Boolean “skins” are often organized as a matrix layout, so it is possible, in addition to detecting a contact, also to determine a smaller zone where such contact occurred.
The patent application US 2015/0073598 A1, Rosenstein et al., entitled “Human Interface Mobile Robot”, in fact, describes a robotized platform called also iRobot (in other publications) covered with a Boolean “skin” (which provides a so-called “touch input”). Said Boolean “skin” implements a tactile interface that allows the operator to move said robotized platform, simply by touching it, and that supports the management procedures of potential collisions that could happen to such platform during its motion. The iRobot platform is therefore just sensitive to a touch, so it is easy to implement a procedure that automatically stops the motion of the platform when it is touched at the side oriented towards the motion; as it is equally easy to implement a procedure that, when the iRobot platform is stationary and it is touched, it starts to move towards the direction which is compatible with the touch (interpreted as a thrust).
The system does not carry out a measurement of the pressure or of the power exerted by means of touches that are detected.
In addition to stop the motion on the occasion of collisions, an important application implemented in the platform iRobot provides that the manual control of the platform is realized with a sort of motorized aid to the motion of the platform itself. In fact, the drive torque generated at the detection of a touch can be dimensioned to be insufficient, even if only slightly, to move the platform; in this way, to move, the platform requires the contribution of a thrust: and precisely the intensity of the thrust, added to the driving torque generated automatically, serves to adjust the speed of displacement.
The significant usefulness of this application can be easily imagined. It allows, for example, to move heavy trolleys with light thrusts, being in fact the fatigue work largely done by the drive torque generated automatically. At the same time, such a trolley, even when it is very heavy, may be preserved from suffering or causing great damages due to accidental collisions, being able to implement, contextually (as mentioned earlier), a prioritized stop procedure in the event of a collision. It should however be noted that the intervention can take place only upon impact, and then the limitation of the damage may not always be sufficient. In general, it can be concluded that the so-called Boolean “skins” represent a technology useful for some applications that do not require particular accuracy of the movement and they can manage, although with distinct procedures, both the control of the movements and some countermeasures in cases of collision.
The analog “skins”, unlike the Boolean ones, allow to estimate the pressure with which a touch is exerted, therefore, they offer an additional control variable which can assume a continuous value within a certain range.
Also the analog “skins” are a technology suitable for the implementation of procedures both for the management of collisions, and for assisted manual handling of the movement. In fact, they allow to implement all that is implementable with Boolean “skins” but, being able to provide information well-richer than the pure detection of the presence or absence of a contact, they can be exploited to improve the performance, in terms of effectiveness and accuracy, which can be reached with the Boolean “skins”.
Said analog “skins” are available in various types and can be realized with quite mature and reliable technology. Typically, they are constituted by conductor or semi-conductor materials manufactured in very thin strips having a variable resistance depending on the pressure that is exerted on them. They are already applied in the field of robotics, but mainly within the constrained mechanical arms, that is, in contexts in which the movements and possible trajectories are strictly limited. Other known applications take place in the field of mechanical manipulation of objects, in which it is important to adjust the pressure of a grip (e.g. in artificial hands, or pliers, of industrial robots). However, there are no known applications in self-moving robotized platforms, where their use is finalized to the control of the ambulation. In particular, in the known applications, the detected pressure is not used as an input variable for the determination of the type of motion that must be actuated. In fact, it is not immediate and easy to conceive a command protocol, based on pressure signals, and able to control correctly, accurately and safely the ambulation of a self-moving robotized platform in a generic environment.
In fact, one of the objectives of each human-machine interface (HMI) is that of allowing a person to issue commands in a natural and spontaneous way. Then the machine must be provided with an interpretative capability, able to understand a wide range of spontaneous and intuitive interactions that can be generated by a human operator.
At this point it matters to keep in mind as the most widespread technologies, sensitive to pressure, and used to make the most common “skins”, enable excellent accuracy for very low pressures, while discriminate more coarsely the highest pressure. This characteristic of sensitivity is a critical factor, but is also a feature that allows to conceive a very accurate method of management (both for manual controls and for collisions). It is a critical factor since both the collisions and the manual thrusts, which may be caused by a human operator, can be excessively strong and, above all, they are manifested as a discontinuous “step” function in which the transition between the absence of pressure and significant pressure happens substantially in an instant. It can be assumed that when a person touches an object, the change of state from “absence-of-contact” to “presence-of-contact” (or vice versa) is a nearly instantaneous transition. The considerable sensitivity to slight pressure, however, is also a useful factor for the implementation of a precise control, since it is also associated with the possibility of discriminating small variations of pressure.
Therefore, the main scope of the present invention is to conceive a new self-moving robotic platform whose mobility can be controlled with adequate accuracy, translating manual user commands into automatically-generated commands for the control of the motion. And said manual user commands can be given in a simple and intuitive way, by exercising a pressure, even of a slight intensity but, at the same time, compatible with the human sensitivity, on the sides of such a platform.
Another object of the present invention is to conceive a protocol, which is also intuitive, which allows to control manually, with the simple exercise of small pressures, the ambulation of a self-moving robotized platform, and such control must be able to be as precise as possible in terms of speed and trajectory, being possible both straight and curved movements, and even pure rotations.
Again, another object of the present invention is to conceive a self-moving robotized platform in which the same system of sensors and actuators used for the control of ambulation also constitutes a system exploitable for the effective management of possible collisions with external elements.
Finally, further object of the present invention is to conceive a self-moving robotized platform in which the rules for the control and the calibration of the system are the most simple and reliable as possible. The intended purposes for this invention are achieved by means of a self-moving robotized platform that is covered on its lateral faces by a cover which serves as a pressure sensor, and said cover which acts as a pressure sensor is constituted by one or more layers that implement at least the following functions:
The main advantage of the present invention is given by the fact that a self-moving robotized platform realized according to the teachings of the present invention, fulfills the main purposes for which it was conceived.
This invention has also further advantages which will be made apparent from the following description, which shows more details, from the attached claims, which form an integral part of the same description, and from the illustration of some examples of embodiment described, but not limited to, in what follows, and in the accompanying drawings in which:
Obviously any type of shape is possible, however, the parallelepiped-shaped platform, in addition to allow a clear illustration of the inventive principles of the present invention, also constitutes one of the preferred shapes as it can be conveniently used as a trolley for the transport of payloads, and can also house all the components necessary for the operation of the same platform. In addition, it can be also suitable to accommodate the integration of eventual mechanisms functional in different applications.
With the number 100 it is then indicated, as a whole, the self-moving robotized platform according to the teachings of the present invention. However, the representation offered in
With the number 101 it is then represented a wheel, it indicates a possible mean of implementing the ambulation of the platform 100, and with the number 102 it is shown a generic load of the platform 100 which can be, as said, a mere load to be transported as well as a device, for example anthropomorphic, which can be an integral part of the same platform, designed to perform a particular function.
Finally, with the number 1-10, some tactile films covering all the side walls of the platform 100 are indicated. Such tactile films 1-10, in the example of
As stated above, it said tactile strips 1-10 constitute a known technology and they can be realized with conductive or semiconductor materials whose resistance is variable depending on the squeezing pressure that is exerted on them. However, such use of tactile strips 1-10, based on the property of those materials whose resistance varies according to the pressure to which they are subjected, is just one of the possible choices for realizing the pressure sensors. In theory, also other materials can be used, in fact a pressure applied on a material can also influence other parameters, such as its capacitive characterization or, more in general, its overall electrical characterization, so that, when said material is inserted in a powered circuit, measurable effects take place, and the pressure exerted on it can be estimated. In other embodiments of the present invention, such pressure sensors can be systems more complex than it is a particular material that varies its electrical behavior as a function of the pressure to which it is subjected. Said pressure sensors can be, for example, real circuits that exploit other characteristics, not strictly electrical, of certain materials (for example of optical behavior characteristics), and such circuits can integrate, inside them, measurement subsystems, and/or other circuit components.
However, for the purposes of this description of the present invention, these pressure sensors can be generically modeled, at a high level, as components of an electrical circuit with an electrical characterization pressure-dependent: it means that a measurement of an electrical quantity can be performed on them, and this measure (e.g. voltage or current levels, or frequency measurements, both continuous and discrete, etc.), which is dependent on the pressure which they are subjected to, is manageable by a computer.
Coming back still to the case taken as an example, in which the tactile strips 1-10 are composed of variable resistance material, the tactile sensor system works when said tactile strips 1-10 are all connected so as to form an electric circuit crossed by current, in which there are inserted measuring elements capable of detecting, selectively, changes in resistance in each of the various tactile strips 110 that substantially cover the entire sides of the self-moving robotized platform 100.
Thus, though not shown in
For completeness it has to be stated that the actuation of the movement of a self-moving robotized platform 100, implemented through a control exercised on the wheels, can be conceived in a wide variety of combinations of drive wheels, steering wheels and free wheels (obviously with different control accuracies): from the simplest case in which there is only one driving wheel, to the most complete case in which all the wheels are driving and steering.
The fact that the tactile strips 1-10 of the example illustrated in
It is clear that it is possible to create a tactile “skin” of said self-moving robotized platform 100 also with other arrangements and forms of the tactile strips 1-10, in order to better discriminate the points where a pressure is detected. Even in the presence of such configurations, and even more in the case of greater precision in the detection of pressure points, the inventive principles that characterize the platform 100, as a whole, are applied and remain valid. It should be said that, when the platform is moving on a substantially flat surface, the configuration of pressure sensors arranged as in
In all three examples [
Even in the examples of
In
In this case, the tactile sensor system detects the pressure 121. And if the platform was very lightweight, it would move with an approximately rectilinear motion in the direction of the thrust. Said pressure 121, detected by the tactile sensor system, can then be intuitively interpreted as a command for the activation of a rectilinear motion, in which acceleration and velocity are calculated instant by instant as a function of the pressure which is applied. The number 131 indicates an arrow which is representative of rectilinear motion that is put in place, for example, through the drive wheels 101.
The calculation of the motion 131 can be made in various ways. One way that certainly can be very intuitive is based on the model of the “empty box”. Applying this model of the “empty box”, the calculation is reduced to simulate the motion that would occur if a pressure equal to that measured was applied to an empty, very lightweight box, having the same size of the robotic platform 100.
Once the speed is determined in this way, by means of an automatic control, the appropriate couples can be applied to the drive wheels so that the instantaneous speed of the self-moving robotic platform 100 corresponds to the instantaneous speed calculated with the model of the “empty box”.
A slight thrust given at the center of a face of the self-moving robotic platform 100, such as that represented in the examples of
At this point, it is possible to infer how to impart other manual intuitive controls, to determine also curvilinear motions, through a slight pressure. In fact, continuing to imagine an “empty box”, and applying a pressure, i.e. a push, no longer at the center of a face, but at a point shifted toward an edge of said face, the resulting motion, in the case of an “empty box”, would be a curved motion which will also tend to rotate this hypothetical “empty box.”
Such a case is represented in the example of
Thus, it is clear that a self-moving robotized platform 100 can be operated manually, making it to follow various types of trajectories, exerting little pressure either in the middle of the side faces, or on more decentralized points. Exerting slight pressures, it is then possible to move the platform 100 very slowly curing the accuracy of movement; while greater accelerations can be imparted by increasing the pressure.
The calculation of the speeds and accelerations which have to be imparted to the platform 100 as a function of the detected pressures is not deepened beyond, since it is evident that multiple strategies may be adopted. The model of the “empty box”, which has been previously mentioned, is very intuitive, and it can also be adopted partially, for example limited to a predetermined range of pressures. The partial use of the model of the “empty box”, in fact, may foresee the insertion of thresholds on the maximum speed or accelerations allowed to the platform.
An important concept can be synthesized and generalized: namely that, on the basis of the information on the extent of pressure manually exerted on a self-moving robotized platform 100, it is possible to conceive the formulas for calculation of the movement commands of said platform 100, and said commands can be intuitive for the operator who have to exert them, and at the same time they may take into account and manage also other application-specific requirements, for example to increase safety.
In
All the examples shown in
It is noted, at this point, as the model of the so-called “empty box” should not necessarily be understood as a model to be strictly reproduced for the definition of the manual control commands. Rather it is just a model to derive inspiration for conceiving a system of manual and intuitive user commands for the movement of a self-moving robotized platform.
In short, the self-moving robotized platform 100 according to the teachings of the present invention can be physically maneuvered by generating automatic control commands that determine the directions of movement and the rotations “compatible” with the manual controls exercised by an operator: in other words, such directions of movement and such rotations are those that the operator can intuitively expect as a consequence of the thrusts he exerts at certain points on the outer surface of the platform 100.
Once the intuitive nature of the controls is preserved, it is certainly possible, and also recommendable, to tune the functions that provide the calculation of the control commands as a function of applied pressure, by means of calibration processes, which can also be empirical.
In addition to the problem of the manual control of the movement, it is noted that the so-described system is a platform which also implements an automatic management of collisions. In fact, when a collision (or impact) happens, the “skin” of the self-moving robotized platform 100 necessarily detects the occurrence of a pressure at a point of its surface. As a consequence of said pressure generated by the impact, the self-moving robotized platform will move in opposition to the movement that caused the bump itself; and consequently the platform 100 will move away from the obstacle bumped. Just detached away from the obstacle, it will no longer sense any pressure, and then it will stop.
The process synthetized above, therefore, would appear to be already a way for the management of the collisions, it would be enough just to state that the commands generated as a result of pressures on the “skin” of the platform 100 must override all other controls; In fact, the platform 100 is typically also controllable with other non-manual commands, for example, it can be controlled remotely.
The above-described management of the bumps, however, presents some obvious drawbacks. In fact, if the bump is just a little violent, the pressure produced by the impact is high and consequently the speed with which the platform 100 is rejected is also high. But then, as soon as the platform detaches from the obstacle, the platform stops, once again rather abruptly due to the total absence of pressure. To avoid such too abrupt behavior, various strategies can be adopted. The expedient adopted according to the teachings of the present invention is illustrated with the support of
The view proposed in
Said cushioning layer 210 can be realized with soft materials, such as foam rubber or particular viscoelastic foams (also called “memory foams”), and it is characterized by an elastic coefficient “k”, indicated in
If the elastic coefficient “k” of the cushioning layer 210 is low enough, during a collision, said cushioning layer 210 is compressed by opposing little resistance, and by transferring to the tactile strip 1-10 a pressure initially very mild, however, enough to instantly generate a command to move the platform 100 in opposition the motion of collision.
Unlike the motion that would be generated as a result of a non-amortized collision, such motion, contrary to the collision, will be soft. Moreover, during said phase of detachment, the cushioning layer 210 must resume its initial shape, and as long as this position is not reached, it continues to exert a pressure, albeit declining, on the tactile strip 1-10, and therefore also the speed of detachment will decrease gradually until the complete separation of the platform 100 from the bumped obstacle.
The great utility of said cushioning layer 210 is therefore evident on the occasion of collisions: it is an important characteristic for the good functioning of the motion control system of the self-moving robotized platforms 100 according to the teachings of the present invention.
Said cushioning layer 210 is of great benefit not just on the occasion of collisions, but it is also useful during the manual operations for the control of the platform. In fact, the typical manual sensitivity of most people is not enough to perform very precise controls. It can happen that, in the moment of contact, a manual push may produce a too sudden and intense pressure, enough to determine a sudden leap forward of the platform 100, and if said leap forward determines a loss of contact, the platform would stop abruptly, then maybe it suddenly starts again, if the hand should restore with a contact, and so on as to produce a start of very irregular motion.
It is possible to choose a compromise by setting quite high thresholds for the pressures able to activate the motion, but considering the fact that the tactile strips 1-10 are typically more sensitive (and accurate) at weak pressures, that choice would give away the possibility to have a precise control of motion.
But, if very low motion activation thresholds were set, to take advantage of the dynamics of sensitivity of the tactile strips 1-10, the manual control of the platform would be allowed only to operators capable of extremely light touch: so it is clear that this limitation would represent an unacceptable usability limit for the entire system.
Even the hypothetical use of sensors sensitive to touches at higher pressures, would have contraindications, since accidental impacts would be managed only when manifestly vigorous. The presence of a cushioning layer 210 is therefore essential to achieve the precision of manual control without limiting it only to people able to act with very lightweight hand. Indeed, a manual touch, filtered by a cushioning layer 210, always produces a slight pressure on the tactile strips 110 as soon as the hand, or the object with which the thrust is exerted, comes in contact with the outer surface of the cushioning layer 210. This gentle pressure slowly starts the motion in the desired direction without causing an unwanted gap between the hand (or the object with which the thrust is exerted) and the platform 100 that has to be handled, and the person who is carrying out the manual control has the time to adjust the pressure with which he intends to push.
A further benefit determined by the presence of said cushioning layer 210 is given by the fact that it distributes the effect of the exerted pressure over a wider surface. So, if the pressure is exerted, at the outer side, in a very reduced area, for example, if the thrust is carried out with a pointed object, the pressure transferred to the tactile strips 1-10 results exerted in a wider area.
Without going into the details of the technology of the tactile strips 1-10, which for the purposes of the present invention can be considered as a known and mature technology, it is important to underline that the precision of a tactile strip 1-10, intended as a pressure sensor, generally improves in the case where the pressure is more distributed.
Anyhow, the interpretation of the output signal coming from the tactile sensor system which constitutes an essential part of the “skin” of the self-moving robotized platform 100, must be calibrated in the final assembly conditions, then with cushioning layer 210 already applied. And the presence of said cushioning layer 210 also facilitates the calibration process by reducing the number of forms of objects with which to exert the calibration thrusts. In a limit case, especially for the calibration of relatively light pressures, only one contact object can be used (perhaps with a form recalling the thrust carried out with a hand).
It should be also pointed out, with regards to the calibration phase, that said cushioning layer 210, can also produce a not useful effect when the control is not performed manually. In fact, sudden accelerations and decelerations (e.g. those imparted with remote control) may cause that, by inertia, said cushioning layer 210 determines slight pressures on the underlying tactile strips 1-10: such pressures, not produced by impulses or impacts, should not be considered for purposes of movement control, and therefore it should be set, in the calibration phase, also a minimum activation pressure threshold. Said minimum activation pressure threshold can be much lower the better said cushioning layer 210 is applied, and if it can be maintained low enough it can reasonably be considered useful, since it allows to avoid the generation of motion even when the platform is just lightly touched.
The motion control system described so far works properly and meets the purposes for which it was designed without the need to synthesize explicitly the piece of information indicating whether a pressure is determined by an accidental collision or by a volunteer thrust. Therefore, it is suited for the implementation of self-moving robotized platforms also extremely simple, where simplicity is certainly an advantage when a satisfactory performance is obtained the same.
However, the control system, as described so far, has enough information to perform a far more elaborate processing that allows to increase the performance both in the ambulation and in the collision control.
In particular, it is possible to provide that the control commands generated by the computing unit, besides being dependent on the sensed instantaneous pressures, also depend on the state of motion of the platform itself.
In fact, a pressure directed against the motion generates a braking, in this case a quite high maximum deceleration may be set, both in the project and during the calibration: this is to allow a sudden braking and to limit the damage from a collision. In the case of a pressure exerted to initiate a motion, instead, lower permissible accelerations should be imposed, so as to have always more sweet starts, also in case of possible abrupt pushes which could be applied unintentionally.
It is clear then, that the functions that bind the generation of the movement commands with the state of motion of the platform 100 may significantly depend on the context of application; and the availability of a self-moving robotized platform 100, designed according to the teachings of the present invention, is well suited to support very flexible implementations of the motion functions: a fine tuning can take place both in the design refinement phase and in in-field calibration phase. A further clarification is about the choice of the “skin” that serves as a pressure sensor. The implementation shown in
However, it is not excluded that both functions can be guaranteed, in the future, with higher quality and at competitive costs, also with different materials that can implement both functions (sensory and cushioning) in an integrated way.
For example, there are known piezo-electric or piezo-resistive materials with which it is possible to realize the pressure sensors, and these materials are also sufficiently deformable so that they can be manufactured with thicknesses suitable to implement also the damping function. At the time, these materials are still a bit too hard, they have not yet adequate sensitivity in the range of pressures of interest for the application of the present invention, and also the costs are not competitive compared to implementation with the bilayer solution.
However, the technological evolution may propose, in the future, materials with characteristics suitable for achieving a “skin” able to serve as a pressure sensor by the use of a single layer made with a material with performance (in terms of quality and cost) better than that currently available. Then it will be possible to resort to a solution of implementation based on a single-layer coverage, without departing from the inventive concepts developed in the present invention. Again, a further implementing option can occur in the case in which particular technologies for implementing the pressure sensors are available. In fact, it is possible to hypothesize a further case in which the layer that implements the function of the sensor is made with materials that confer elasticity and flexibility, consequently this layer can be deformed, even significantly, when subjected to a pressure, and then return to its initial shape.
A pressure sensor of this type can therefore perform, to a certain extent, also the cushioning function. However, unlike in the previously illustrated variant of implementation, in which the possibility of having a cover formed with a single layer is mentioned, in many cases the elastic characteristics and the thicknesses of these layers are not suitable for the realization of a tactile cover able to implement a cushioning function sufficient for the purposes of the present invention. In such cases it is therefore necessary to increase the cushioning function by the addition of a further layer that performs a purely cushioning function, which can be typically placed below the layer that implements the function of the sensor.
Therefore, generally speaking, the implementation of the cover of the self-moving robotized platform (100) which serves as a pressure sensor can also be implemented according to a further variant with respect to the preferred implementation illustrated in
Unlike the rectilinear motion 131 exemplified in
The case of
Unlike the case of
The processing of data about the pressures exerted on the “skin” of the platform 100, however, not only allows to calculate a resultant of the applied forces, but also allows to detect the extent of the forces which, while exerted on the sides of the platform, are opposed canceling each other.
The case in which the entity of the forces which cancel, tightening the platform 100, is decidedly prevailing with respect to the possible formation of small torques, corresponds to a sort of intention of grabbing the platform 100, by the operator who is manipulating it.
And, therefore, such intention of grabbing the platform 100 can be synthesized through the processing of the data from the “skin” (it is recalled that it is nothing else than a system of analog touch sensors, sensitive to pressure), and the cases in which two opposing forces are exerted roughly on the same line can be identified: approximations are necessary because the evaluations have to do with the interpretation of manual and spontaneous controls. In order to preserve the prerogative that imposes just manual controls designed to be intuitive, it is then appropriate to consider the event that an operator, acting with spontaneity, and wanting to move the platform 100 by pulling it toward him, tightens it between his hands and pulls.
The interpretation of such behavior, would require a “skin” composed by tactile sensors able to measure not just the pressure but also the components of force parallel to the surface. It is theoretically possible to use sensors of this type, but they do not constitute a low-cost technology and easy to be supplied (they are mostly laboratory material), and furthermore, even if available, their exploitation would require a considerable complexity, even circuital.
Although not excluding, in principle, the possibility of using such advanced “skins” also sensitive to the forces exerted by friction and having components parallel to the surfaces covered by such “skins”, the system implemented according to the teachings of the present invention can support the interpretation of a manual command imparted as depicted in
However, in the case in which these techniques based on the analysis of the torques generated on the wheels are used in combination with the techniques based on the detection of the contacts on the surface of the platform 100, the drawbacks can be significantly limited, as the analysis of the torques on the wheels can be considered, and used, to derive only a specific component of the motion desired by the operator, who may continue to dictate the other components of the motion by the aid of the tactile sensor system.
For example, rotations can be obtained by adjusting the pressure of the grip in points not perfectly opposite, thus giving rise to a resultant of the pressures which also provides a torque, or an oblique motion can be obtained by exerting a different pressure with the two hands, thus only the pressure exerted by one side vanishes, while the pressure exerted on the opposite side is not completely compensated.
The cited example is then generalizable, and indicates a further possibility of control offered by the self-moving robotized platform 100 made according to the teachings of the present invention. Such command opportunity allows to preserve the spontaneous and typical behavior that an operator can put in place when it intends to manually maneuver the movement of the platform 100. And this possibility is given by the fact that the movement to be imparted to the self-moving robotized platform 100 can be calculated not only by determining the resultant of the pressures applied on the surface; in fact, the analysis of these pressures, when exercised in more points, can also reveal the presence of grabs, namely the presence of components of forces that oppose without necessarily generate significant torques (so rotations), and in the presence of such grabs can therefore provide other information for the analysis, such as the occurrence of torques applied on at least one wheel of the platform. The occurrence of such a type of torque, applied on at least one wheel, allows to detect intention of operator of imparting a component of motion in the direction perpendicular to the force components which vanish each other by opposing.
And so, it is then possible to transform the identification of this intention into a corresponding movement command.
The analysis of torques present on the wheels is also useful as it allows to distinguish the case in which an operator pushes (intentionally or for distraction) the platform against an obstacle. In fact, also in this case, the tactile sensor system detects opposite forces that cancel each other: the thrust exerted by the operator and the opposite resisting force from the obstacle. However, in such cases, it doesn't arise on the wheels a torque corresponding to a movement perpendicular to the force components which are opposed, and the case can thus be managed and treated differently.
Depending on the complexity of the processing, and on the accuracy of the data coming from the measurement of the torque variations on the wheels, a more or less precise control can be achieved. However, the compromise in precision is certainly acceptable with respect to the achievable result, which is to devise a system capable of interpreting with completeness a great variety of spontaneous manual controls that can be given by an operator.
Even in the case that some of the platform motions are imparted as a result of these “gripping” command, the system can be calibrated once assembled, this calibration phase can be designed so as to take account of the typical use cases, and in this way also the precision achievable with this type of command can be further improved.
It has been underlined in many parts of this description that the leading goal that has guided the implementation of the commands for the control of the motion of the self-moving robotized platform according to the teachings of the present invention, was to conceive an intuitive set of control commands that enable an operator, even after a very short period of practice, to manually operate the platform.
This objective can be achieved, as mentioned above, inspired by a very intuitive model, such as the model of the “empty box.” It means that, the operator has the ability to impart the movement control commands simply imagining to push or pull a lightweight box, while, in many cases, it is actually a platform having considerable weight and inertias.
Well, the self-moving robotized platform according to the present description (i.e. provided with a tactile “skin” sensitive to pressure and associated to appropriate processing means suitable to generate commands for the movement of the platform itself) can support the generation of commands which may be inspired to other intuitive models. In fact, the “empty box” might be imagined to be “sticky” too, as well as lightweight; having surface completely “sticky”, or “sticky” only in some parts appropriately identified. It is therefore possible to refer to a model, intuitive again, such as the model of the “sticky empty box.”
Such model is intuitive as it is immediate to imagine what can happen by touching a lightweight box, but very “sticky.” In this case, after the touch with a hand, it would remain stuck to the box, and withdrawing said hand quite slowly, the box would continue to remain attached to the hand, and then, being said box actually light, it would be pulled, even in the absence of handles.
Otherwise, if the hand was withdrawn with a sudden movement it probably would detach without pulling the box with continuity.
Then it is obvious to think that, if the hand continues to push, the box would move in the direction of the thrust, exactly as in the cases discussed extensively in the previous part of this description.
The intuitive behavior described above, which regards a “sticky empty box” can be easily implemented, with more or less approximation, in the platform conceived in accordance with the present invention.
In fact, since the various pressures exerted on the side surfaces of the platform are known as well as the pressure points (for the presence of the tactile “skin”), it is also possible to determine whether or not a pressure is exerted in a “sticky point”. In this case, in addition to the control commands generated as explained above, consistent with the model of the simple “empty box”, additional commands must also be generated in correspondence to the changes of pressure that are determined by the movement of withdrawal of the hand (or of the object with which the platform had been touched). The commands, which have to be generated in this case, control the motion of the platform in order to determine a movement similar to the one which would occur if the platform were pulled from the contact point (sticky).
Slow movements of retraction allow the platform to follow the hand (or the object used to touch the platform) maintaining a contact and preserving a light pressure. Then, if the movements allowed to the platform implemented in this way are quite smooth, by placing a limit on the permitted speed and acceleration, an eventual abrupt retraction movement of the hand (or the object used to touch the platform) could not be followed by the platform, there would be the detachment and the platform would no longer be pulled thanks to the so-called “sticky” effect.
The mode of detachment indicated above is absolutely intuitive, and it is also easily implementable in the self-moving platform realized in accordance with the present invention; nevertheless, there are other ways, just as easily implementable and equally intuitive. For example, an operator may detach the hand by touching the platform at another point to hold it: it is evident that, being possible to detect also this further touch, it is easy to interpret the will of detachment from the operator.
Definitively, the intuitive “sticky empty box” effect is easily deployable and also useful in a number of circumstances where it is not easy neither push in the right direction nor even grabbing the platform. For example, in cases where a platform is parked in a sort of niche (in which it was uncomfortable to grab and pull it out), it may be particularly convenient, just to touch the platform, remain “stuck”, and pull it out slowly. The implementation of this “sticky empty box” effect is, as already said, very simple. It is sufficient to determine whether a pressure is exerted at a point to be considered “sticky”, and to interpret appropriately the pressure variations in the sense of their reduction, as indicators of a command of pulling the platform (in the case of slow movements) or as indicators of release intention (fast movements). In general, particular pressures on the self-moving platform (100), defined by their point of application, by their intensity or variation of intensity, can thus be associated with movements of the platform in the direction towards the operator; the operator can interpret these movements intuitively by making a more or less faithful analogy with the behavior of a lightweight object that remains “stuck” to him in the contact point.
The points, or the circumstances, in which the lateral surface of the platform is “sticky” can be determined by many variants of implementation. Among these, few example may be cited: the simplest is the definition of permanent “sticky” areas (possibly highlighted graphically on the side surfaces of the platform), or a “sticky” area may be activated on request by additional commands (e.g. a voice command).
Also the detachment can occur according to many variants. In addition to the examples already mentioned with abrupt movement, or by retaining the platform with the other hand, it is also possible to imagine other possibilities, based on commands of another type (e.g. a voice command) with which to inhibit the “stickiness”, or with other types of movement, such as a swiping movement, which, moreover, would be fairly intuitive (in fact, you can detach from a “sticky” surface swiping over it). A further option of implementation, which may be useful in many applications, and that therefore it is worth to mention explicitly, is to conceive of a tactile “skin” sensitive to pressure, but removable and applicable on different platforms or even (when convenient) directly on the payload that is placed on the platform itself. An example well illustrates this option of implementation.
It is then considered, but not limited to, the example of a self-moving cart controlled by a more or less remote control. Continuing to remaining within the considered example, for certain uses, it is convenient that said self-moving cart is provided with the possibility to be controlled, as well as remotely, also manually, and in an intuitive manner, by an operator. In this case, a wrapper constituted by a removable tactile “skin”, sensitive to pressure, as described in the present invention, can be applied on said self-moving cart when required.
For example, said removable tactile “skin” can be applied so as to surround the platform (possibly already loaded) along its perimeter, and placed in such a position that covers the surfaces in the zones on which is comfortable to manually push the cart. Said removable tactile “skin”, applied so as to be well-fitting, would work exactly as in previously described cases (cases in which the said tactile “skin” sensitive to the pressure is instead permanently applied on the self-moving platform), and would provide the pressure information that are necessary (and sufficient) to generate the control commands for the platform movement, as previously illustrated for the case of fixed tactile “skins”.
Such pressure information should be processed appropriately, not for the direct generation of the control commands for the movement, but rather to generate intermediate commands; for example, commands to be transmitted to the existing remote control, so that the latter generates the control commands for the movement of the platform, according to the intuitive model which is adopted (“empty box”, or “sticky empty box”, or others).
Therefore, by synthesizing the just exposed variant, in order to catch a generalization from the foregoing by way of example, it can be concluded that generic self-moving platforms can be equipped, even temporarily, with a removable tactile “skin” sensitive to the pressure, so that they are, however, self-moving platforms according to the teachings of the present invention.
In order to allow the overall system to work properly, it is obviously useful to devise a removable system that also includes, in addition to the removable tactile “skin”, appropriate computing means and interface means suitable to support communications and data exchanges. Said computing means perform an initial processing of the data provided by the system of the sensors of the tactile “skin”; while said interface means are suitable to be coupled to other control systems. In fact, the application of a removable tactile “skin” requires to integrate the processing of the pressure data detected by said tactile removable “skin” into the procedures of the control systems for the movement of the considered self-moving platform (control systems which are typically already present in the platform). Therefore, these removable tactile “skins” sensitive to the pressure must always be associated with the appropriate “drivers” specific for the platform on which they are applied, so as to adapt the processes executed at the level of “skins” with the control system of the platform on which they are applied. Obviously, many of the properties listed to describe the option based on the removable tactile “skin” are also found in implementations with permanent tactile “skin”. In particular, the decoupling between the processing of the data coming from the pressure sensors (according to the interpretation of an intuitive model, e.g. the model of the “empty box”) and the generation of the control commands on the motors of the wheel, is an implementation choice which can also be adopted in self-moving platforms (100) with permanent tactile “skins”.
However, such a definite decoupling, which also provides for the use of distinct computing means and needs a suitable physical interface, is a typical option for implementations based on removable tactile “skins”.
What is important to point out is the fact that the availability of continuous information about the extent of the contact pressure, allows an effective integration of control methodologies also based on many sensors, by performing the coherent interpretations of the operator's will, and by distinguishing the different cases, which is often not possible by the use of mere Boolean sensors of contact.
In general, it is noted that an interesting prerogative of the present invention is that of being easily integrable with other complementary solutions, in order to translate the commands given by the user manually or intuitively into the motion control commands of a self-moving robotized platform 100.
For example, if the platform has handles, protruding edges, or other points that are suitable to be grasped, or in general utilized by an operator to intuitively control the movements of such platform, these parts of the platform may be equipped with suitable sensors which detect these actions by the operator (typically other pressure sensors), and the detected signals can be used to elaborate the corresponding automatic control commands of movement.
It is clear that the number of forms in which these additional gripping points can be implemented is extremely diverse, each can have unique characteristics and may or may not be relevant to the management of collisions, and therefore the measurement data must be interpreted and managed in a specific way for each case. What matters to highlight here is the fact that even these options, useful to improve the overall usability of the final implementation of a self-moving robotized platform 100, as well as useful, are also very easily integrable in a self-moving robotized platform 100 that is designed according to the teachings of the present invention.
The invention just described can also lend itself to numerous variations that may offer additional advantages compared to those previously mentioned. And these further variants can be developed by the man skilled in the art without departing from the invention as it is clear from the present description and the claims appended hereto. Therefore, the arrangement of some described elements can be changed; and each element can be developed in different materials, size or shape; furthermore, the invention may be implemented in a partial way as well as other described details may be replaced by technically and functionally equivalent elements.
In particular, and as already said above, the particular technologies concerning the use of specific materials is not a characterizing issue of the present invention and therefore, if were to be available in the future more advantageous materials than those indicated in preferred implementations, or if the current technologies relating to the sensors and to the numerical control evolve towards better performance or best quality-price ratios, so as to apply more efficiently the different controls required, further improvements could be made in the embodiment of the present invention without in any way changing the inventive nature.
Finally, the described invention may incorporate and support further measures aimed to further improve the performance and the functions that can be implemented by a self-moving robotized platform whose motion can be adequately controlled: these measures, not described here, may eventually be the subject of further patent applications associable to this invention.
| Number | Date | Country | Kind |
|---|---|---|---|
| IT 10201500007252 | Nov 2015 | IT | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/IB2016/056707 | 11/8/2016 | WO | 00 |
