REDUCTION OF CONTACT FORCE OVER-SWIMMING IN PNEUMATIC FORCE CONTROL DEVICES

Abstract

Method for controlling a robot-assisted force control device (10) comprising a robot-assisted tool (22), comprising: in a position control mode, positioning the robot-assisted tool (22) along a compliance stroke of the force control device (10), away from a mechanical stop in the force control device (10); in the position control mode, moving the robot-assisted tool (22) and / or a workpiece (24) towards each other; in the position control mode, detection of contact between the robot-assisted tool (22) and the workpiece (24), wherein a force exerted by the robot-assisted tool (22) on the workpiece (24) upon contact is less than a force setpoint; Performing a smooth transition from position control mode to force control mode; and in the force control mode, increasing the force exerted by the robot-assisted tool (22) on the workpiece (24) to the target force value.

Description

[0001] The present disclosure relates generally to robotics, and more particularly to reducing force overshoot in pneumatic robotic force controllers upon contact with a workpiece. BACKGROUND

[0002] Robots are an indispensable component in the manufacturing, testing, assembly, and packaging of products, in surgical and telesurgical aids, in space exploration, in operations in hazardous environments, and in many other applications. Many robots or tools attached to them measure forces (including torques) applied by contact with a workpiece or the environment, or experienced by the robot as a result of such contact. These forces are fed back to the robot's controller, which controls the robot's position and / or motion based at least in part on the force control. Such systems are known as force control devices. Examples include material removal (sanding, grinding, and the like), automatic part placement, remote excavation or other manipulation of the environment, and the like.Without loss of generality, the present disclosure considers material removal tools as a representative case of the need for and design of robotic force control devices.

[0003] In many manufacturing processes, the basic dimensions of parts are achieved through machining, casting, forging, molding, or similar processes. These processes are sufficient to meet dimensional specifications, but the parts require additional finishing to achieve the desired surface finish. For example, machined parts may require the removal of residual marks and jagged edges. Another example: injection-molded, cast, or forged parts may have burrs, gates, and / or parting lines that must be removed. Robotic tooling for material removal is often used for these finishing operations.

[0004] For a variety of reasons, such as the rigidity of a robot arm, the achievable granularity of its robot-assisted positioning, and the like, real-time force adjustments are difficult or impossible to achieve during real-world material removal operations. Accordingly, compliant force controllers are known in the art. A force controller is an arrangement between the robot arm and a material removal tool (grinder, grinder, etc.) that allows a certain amount of compliance or movement of the material removal tool relative to the robot arm—also referred to as "compliance" or "slippage." A compliant force controller allows the material removal tool to follow slight contour changes of a workpiece while maintaining the programmed robot path in a straight line.A well-known mechanism for providing compliance in force control is a pneumatic cylinder.

[0005] In some respects, the state of the art in robotic material removal still strives to imitate human performance. Fig. Figure 1 shows the measured force exerted by a human operator on a workpiece during material removal. The curve on the left shows the force applied when applying rotary motions with handheld sandpaper. The curve on the right shows the force applied during linear motions with sandpaper in a sanding block. As these curves demonstrate, the operator can easily regulate the increase in applied force at the start of material removal.

[0006] The initial contact of a robotic material removal tool with a workpiece - even through a compliant force control device - typically exhibits an overshoot of the impact force, which can result in poor surface quality or even damage to the workpiece.

[0007] The "Background" section of this document is intended to place aspects of the present disclosure in a technological and operational context to assist those skilled in the art in understanding its scope and utility. Unless expressly identified as such, no statement in this document is admitted to be prior art merely by its inclusion in the "Background" section.

[0008] A method for controlling a robot-assisted force control device and / or a force control device according to the prior art are known from DE 10 2021 106 990 A1, DE 10 2019 105 022 B3, DE 10 2018 106 086 A1, DE 10 2015 104 164 B4, DE 10 2011 006 679 B4, DE 10 2010 003 697 B4.

[0009] The object of the invention is to develop methods for controlling a robot-assisted force control device and / or a force control device which optimizes surface processing and prevents damage through improved force adaptation and upon contact with a workpiece. SUMMARY

[0010] The aforementioned problems are solved by the invention according to the patent claims. The dependent claims define further advantageous embodiments of the invention.

[0011] The following is a simplified summary of the disclosure to provide a basic understanding to those skilled in the art. This summary is not a comprehensive overview of the disclosure and is not intended to identify essential / critical elements of aspects of the disclosure or to delimit the scope of the disclosure. The sole purpose of this summary is to present some concepts disclosed herein in a simplified form as a prelude to the more detailed description presented later.

[0012] According to one or more aspects described and claimed herein, a force control device mitigates or eliminates an overshoot in the impact force upon contact between a robotic tool and a workpiece. The contact is detected while the force control device is operating in a position control mode, using either a steady-state search method or a transient search method. The force applied to the workpiece at contact is less than a predetermined target force. When contact is detected, the force control device smoothly transitions to a force control mode. In the force control mode, the force control device ramps the contact force to the predetermined target force. The force ramp can be linear or along a user-defined trajectory.The stiffness of the force control device is different in position control and force control modes and is controlled by the back pressure in a pneumatic cylinder.

[0013] One aspect relates to a method for controlling a robotic force control device comprising a robotic tool. In a position control mode, the robotic tool is positioned along a compliance stroke of the force control device away from a mechanical stop in the force control device. The robotic tool and / or a workpiece are moved toward each other. Contact between the robotic tool and the workpiece is detected. A smooth transition from the position control mode to the force control mode occurs. In the force control mode, a force exerted by the robotic tool on the workpiece is increased to a force setpoint. A force exerted by the robotic tool on the workpiece upon contact is less than the force setpoint.

[0014] Another aspect relates to a force control device having a compliance motion and configured to hold a robotic tool and apply a desired force to a workpiece by the robotic tool. The force control device includes a controller operable in a position control mode to position the robotic tool along a compliance stroke of the force control device, away from a mechanical stop in the force control device; and, as the robotic tool and / or the workpiece move toward each other, to detect contact between the robotic tool and the workpiece. The controller is further operable in a force control mode to increase a force applied by the robotic tool to the workpiece to a force setpoint.The controller is designed to perform a smooth transition from the position control mode to a force control mode. The force exerted by the robot-assisted tool on the workpiece upon contact is less than the force setpoint. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present disclosure is described in more detail below with reference to the accompanying drawings, in which aspects of the disclosure are illustrated. However, this disclosure should not be construed as limited to the aspects set forth herein. Rather, these aspects are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. Like reference numerals refer to like elements throughout. Fig.Figure 1 is a graph of the force applied by a human operator to a workpiece during material removal operations. Fig. Figure 2 is a diagram of a compliant force control device for a robotic material removal tool. Fig. Figure 3A is a diagram of the approach of a robotic material removal tool to a workpiece. Fig. Figure 3B is a diagram of initial contact between the robotic material removal tool and the workpiece. Fig. Figure 3C is a diagram of stationary or sustained contact between the robotic material removal tool and the workpiece. Fig. Figure 4 is a graph of the force exerted by the robotic material removal tool over the Fig. 3A-C shown positions on the workpiece. Fig.Figure 5 is a graph of force control carriage speed versus time during initial contact between a robotic material removal tool and a workpiece. Fig. Figure 6 shows the initial and stationary position of a piston in a pneumatic cylinder. Fig. Figure 7A is a graph of force versus time for an initial contact with a workpiece for a low airflow force control device to reduce force overshoot. Fig. Figure 7B is a graph of pressure in two chambers versus time according to the Fig. 7A shown force. Fig. Figure 8A is a force versus time graph for initial contact with a workpiece for a high airflow force control device to reduce force overshoots. Fig.Figure 8B is a graph of pressure in two chambers versus time according to the Fig. 8A shown force. In Fig. Figure 9A shows a robotic tool prior to contact with a workpiece. Fig. Figure 9B shows the robot-assisted tool in contact with the workpiece. In Fig. 10 shows some parameters of a force control device. Fig. 11 is a diagram showing the positions of a reference point in a force control device in a method of searching for a steady state for contact detection. Fig. 12A is a timing chart showing contact of a robotic tool with a workpiece in the steady-state search method for contact detection. Fig.12B is a timing chart showing the speed of a robotic tool before, during, and after contact with the workpiece in the steady-state search method for contact detection. Fig. 12C is a timing chart showing changes in a reference position of a force controller before, during, and after contact of the robot tool with the workpiece in the steady-state search method for contact detection. Fig. Figure 12D is a timing diagram showing the force applied by a robotic tool to a workpiece before, during, and after contact in the steady-state search method for contact detection. Fig. 13 is a diagram illustrating the positions of a reference point in a force control device in a method of searching for a transition for contact detection. Fig. 14A is a timing diagram showing contact of a robotic tool with a workpiece in the contact detection transition search method. Fig. Figure 14B is a timing diagram showing the speed of a robotic tool before, during, and after contact with the workpiece in the contact detection transition search process. Fig. 14C is a timing chart showing changes in a reference position of a force controller before, during, and after contact of the robot tool with the workpiece in the contact detection transition search process. Fig. Figure 14D is a timing diagram showing the force applied by a robotic tool to a workpiece before, during, and after contact in the contact detection transition search method. Fig.15 is a control system diagram of a force control device. Fig. Figure 16 is a graph of a force exerted by a robotic tool on a workpiece at and after contact. Fig. 17 is a flowchart of the steps of a method for controlling a robotic force control device. Fig. 18 is a pneumatic schematic diagram of a first aspect of system stiffness control. Fig. 19 is a pneumatic schematic diagram of a second aspect of system stiffness control. DETAILED DESCRIPTION

[0016] For simplicity and illustration, the present disclosure will be described primarily with reference to one exemplary aspect. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present disclosure. However, it will be readily apparent to one skilled in the art that the present disclosure may be practiced without limitation to these specific details. In this description, well-known methods and structures are not described in detail in order not to unnecessarily obscure the present disclosure.

[0017] Fig.Figure 2 shows a simplified view of a representative compliant robotic force control device (FCD) 10, which includes a housing 12, a pneumatic cylinder 14 with a pneumatic piston 16 and a rod 18, and a linear carriage 20 connected to the rod 18. A material removal tool 22 is connected to the carriage 20. When the material removal tool 22 engages a workpiece 24, it can move linearly along a normal axis, as indicated by the motion arrow. The carriage 20, and thus the robotic tool 22, can move linearly in both directions by an amount referred to herein as the "stroke" of the FCD 10. At least the forward extent of the stroke is defined by a mechanical stop.

[0018] A pneumatic control system 26 regulates the force exerted by the material removal tool 22 on the workpiece 24 and ensures a compliant movement of the material removal tool 22 by regulating or controlling the pressure in the chambers of the cylinder 14 on both sides of the piston 16.

[0019] The impact force overshoot analysis is based on considering the position of a material removal tool 22 before, during, and after initial contact with a workpiece 24. It is known that elastic deformations of the tool or workpiece surface can contribute to reducing impact energy. The analysis assumes an impact condition where the tool and workpiece are considered rigid, thus encompassing a worst-case contact event.

[0020] Fig.Figure 3A shows the robotic material removal tool 22 approaching a workpiece 24 at a relative velocity v. This is known as process-to-part (workpiece) motion and is assumed here. Of course, the same analysis also applies to part-to-process motion, where the tool 22 is stationary and the workpiece 24 is moved into contact with it. Prior to contact, the piston 16 in the pneumatic cylinder 14 is generally positioned forward, allowing for compliance movement of the material removal tool 22.

[0021] Fig. Figure 3B shows the initial contact between the material removal tool 22 and the workpiece 24, generating an equal and opposite applied force and a reaction force. Although the material removal tool 22 stops at this point, the relative movement of the force control device housing 12 continues.

[0022] Fig.Figure 3C shows a continuous or stationary contact between the material removal tool 22 and the workpiece 24. It should be noted that the housing 12 of the force control device is displaced from its position at the initial contact ( Fig. 3B) is displaced by a distance d; the piston 16 is displaced by the same distance within the pneumatic cylinder 14. The relative speed between the force control device 10 and the workpiece 24 is now zero.

[0023] In Fig. 4 is the force between the material removal tool 22 and the workpiece 24 when their relative positions are changed by the force Fig. 3A-C are graphically depicted. At the beginning, during the approach ( Fig. 3A), the force is zero. Immediately after the first contact, a force peak occurs ( Fig.3B). The force peak comprises the desired continuous or steady-state force (also referred to as the target force or force setpoint) and a force overshoot. The force overshoot is often detrimental and always undesirable.

[0024] The force overshoot results primarily from two force sources. The first results from the dynamics of the force control device 10 and is modeled by equation (1): Force control device = mdvdt−βv

[0025] The first term in equation (1) represents the change in linear momentum of the force control device housing 12 when the material removal tool 22 contacts the workpiece 24. Fig. Figure 5 shows the rapid speed changes of the housing of the force control device 12 relative to the carriage 20 and to the tool 22 during the duration of the force overshoot.

[0026] The second term in equation (1) refers to a frictional force caused by the movement of the force control device's housing 12 and the viscous friction of moving components. Since most force control devices 10 strive to minimize friction in the system to maximize various performance aspects, the first term in equation (1) dominates, and the second term can be neglected.

[0027] The second cause of the force overshoot is the pressure dynamics in the pneumatic cylinder 14. Fig. Figure 6 illustrates the pressure changes resulting from the volume changes in the chambers, which drives this dynamic. At the time of first contact ( Fig.3B), the slide 20 is positioned so that the cylinder piston 16 is in position x1. After contact of the material removal tool 22 with the workpiece 24, the relative movement difference between the workpiece 22 and the housing 12 of the force control device drives the slide 20 to a set position x2. The change in the position of the slide 20 and thus the stroke of the cylinder piston 16 is dx. A simple force balance across the cylinder 14 shows the output force of the cylinder 12: Fcylinder=P1A1−P2A2

[0028] At the time of the first contact (Initial Contact (IC)) the pressures P 1_IC and P 2_IC set so that F cylinder corresponds to the desired contact force. Furthermore, the force control device 10 assumes the stroke end position to seek contact with the workpiece. This position is assigned to the cylinder volumes V 1_IC and V 2_ICin the chambers of cylinder 14 formed by piston 16. During the initial contact phase, cylinder piston 16 is displaced by a distance dx. Assuming that compressed air is an ideal gas and the displacement occurs sufficiently quickly, the process can be approximately described as adiabatic. The change in the force delivered by cylinder 14 from initial contact to sustained contact (Steady State Contact (SSC)) is given in Equation 3: ΔFcylinder=PI_IC(VI_ICVI_IC+dxA1)kA1−P2_IC(V2_ICV2_IC−dxA2)kA2 where k is the ratio of the specific heat and for air k=1.4.

[0029] As equation 3 shows, the displacement of piston 16 to compress chamber 1 results in an increase in the pressure in chamber 1. Similarly, the displacement of piston 16 to increase the volume in chamber 2 results in a decrease in the pressure in chamber 2.

[0030] The force change described in equation (3) depends on the assumption that the displacement of the piston 16 by dx occurs over a short period of time dt. The duration of dt depends on how fast air can flow into and out of the chambers of the cylinder 14. If the air mass flow into and out of the chambers is slow, the effect occurs at larger values ​​of dt (i.e., at slower relative impact velocities). If the air mass flow is high, the effect is strongest at smaller values ​​of dt (i.e., larger relative impact velocities). The strength of the effect of the pressure change on the force overshoot is shown in the Fig. 7A-B and Fig. 8A-B, in which force ( Fig. 7A, Fig. 8A) and pressure ( Fig. 7B, Fig. 8B) are plotted against time. Fig. Figure 7 shows a slower air flow and illustrates how dramatic this effect can be. Fig.Figure 8 shows a faster air flow and shows that the effect is smaller but still present.

[0031] Although the above analysis focused on contact-phase forces, other perturbing forces that cause deviation from the desired force typically act through the same mechanisms. Therefore, the force overshoot mitigation methods disclosed here are generally applicable to both contact-phase forces and process perturbation forces.

[0032] A known method for avoiding force overshoot during contact between a robot-assisted tool 22 and a workpiece 24 is to lower the force setpoint during the initial contact phase. This ensures that the force overshoot during contact does not exceed the desired force setpoint. Once contact is detected, the force setpoint is increased until it reaches the force required to machine the workpiece 24.

[0033] There are three important aspects to this approach. First, the initial "SearchPoint" and "SearchPosition" must be defined. These are the initial setpoint for the impact force and the position of the slide 20, respectively. Second, there must be a method for determining tool / workpiece contact and loss of contact events. Third, a defined response trajectory is required to achieve the desired process force once a contact event is detected.

[0034] The motion primitives of a system with one degree of freedom (1-DOF) allow to analyze the contact between two rigid bodies by a simple distance function, as described in the Fig. 9A and Fig. 9B. In Fig. 9A, the distance d between the robot-assisted tool 22 and the workpiece 24 is greater than zero. The contact is defined as the Fig.9B, where d=0. If a first-order model of the contact is defined, then a stable contact is defined if d=0 and d˙=0 where d is the first derivative of distance with respect to time or the velocity.

[0035] The key to stable contact is therefore zero velocity, in addition to zero distance function itself. Unfortunately, although the distance function is defined relative to the two objects whose contact is the object of interest, this distance function is rarely actually known in robotic applications. The available information is usually an estimate of the reference position of one or both objects. This means that velocity information has no real relevance for contact detection, since the relative motion between the bodies is unknown.

[0036] Since a general solution is limited by the impossibility of obtaining the distance function, more specific solutions are disclosed here. This analysis begins with the state information for the directly controlled object—in this case, a force control device 10 holding a robotic tool 22. Fig. Figure 10 shows a typical setup of a compliant tool (without the tool), highlighting the feedback terms for force, position, and velocity. Since the absolute motion of each object is unknown, contact is detected by defining a state for one or more of the controlled system variables. By monitoring a deviation of the controlled parameter(s), a contact event can be accurately and reliably detected.

[0037] According to aspects of the present disclosure, a robotic force controller (FCD) 10 is configured to be inserted between a robot arm and a robotic tool 22. In one aspect, the robotic FCD 10 may be configured as shown in Fig.1. In this configuration, the robotic FCD 10 includes a carriage 20 capable of linear movement along an axis and configured to be connected to a robotic tool 24. The robotic FCD 10 further includes a pneumatic cylinder 14 disposed within a housing 12. The cylinder 14 includes a piston 16 dividing the cylinder into first and second chambers. A rod 18 is connected to the piston 16 and extends from the cylinder 14. The rod 18 is connected to the carriage 20 external to the cylinder 14. The robotic FCD 10 further includes a pneumatic control system 26 in airflow communication with the first and second cylinder chambers via first and second air lines, respectively.As is known in the art, before (and after) contact between the robot-assisted tool 22 and a workpiece 24, the position of the piston 16 within the cylinder 14—and thus the position of the carriage 20 along its range of travel or stroke—is controlled by adjusting the pneumatic pressures in the first and second cylinder chambers. After contact with a workpiece 24, the force setpoint—that is, the force with which the robot-assisted tool 22 presses against the workpiece 24—is also controlled via pneumatic pressures. A force measuring device (in . Fig. 1 not shown) monitors this force. The FCD 10 of the Fig.1 is merely a representative example. As will be appreciated by those skilled in the art, an FCD 10 may be constructed and configured in a variety of ways. Generally, the FCD 10 controls the position and / or force of a robotic tool along a compliance stroke, or a distance the FCD 10 can move the robotic tool independently of the movement of a robot arm.

[0038] According to a first aspect of the present disclosure, a steady state search method is used to detect contact between a robotic tool 22 and a workpiece 24. The force control device 10 is as shown in Fig.11. The FCD 10 moves the robotic tool 22 between mechanical stops at both ends of its compliance range of motion or stroke. To prevent the robotic tool 22 from being positioned at either end of the FCD 10's stroke, positional tolerances or exclusion zones are programmed into the FCD 10's motion control software. For example, for an FCD 10 capable of controlling its position to an accuracy of 0.3 mm, the positional tolerances may prevent the robotic tool 22 from being positioned within 0.5 mm of a mechanical stop at either end of its stroke. An initial seek position, preferably facing forward (toward the workpiece 24), is selected as half the stroke of the FCD 10. In some cases, the seek position may be user-defined.As is known in the art, the pressures in the chambers in front of and behind the piston 16 are adjusted to position the robotic tool 22 at the selected search position and then balanced so that the net force F net = 0.

[0039] The robot arm is then slowly moved toward the workpiece 24 (or vice versa). In one aspect, contact between the robot-supported tool 22 and the workpiece 24 is detected by detecting a change in the position of the robot-supported tool 22 along the compliance travel of the FCD 10 opposite to the desired force vector. In another aspect, contact between the robot-supported tool 22 and the workpiece 24 is detected by detecting a contact force that exceeds a predetermined threshold. In response to detecting the contact, movement of the robot arm is controlled to bring the carriage to the final working position, as shown in Fig.11. Loss of contact as the robot arm moves away from the workpiece 24 is detected by the position of the robotic tool 22 being within the forward position tolerance of the FCD 10 (i.e., within or touching the threshold of the front mechanical stop).

[0040] Fig. 12A shows the actual contact, and Fig. 12B-C show the speed, position, and force of the robotic tool 22 for the steady-state search process. When the robotic tool 22 transitions from non-contacting the workpiece 24 to contacting ( Fig. 12A), it initially moves in a direction opposite to the desired force ( Fig. 12B, Fig. 12C). This change in position of the robot-assisted tool 22 or alternatively a detected increased force ( Fig.12D) signals contact with the workpiece 24, which can trigger a stop in the forward movement of the robot arm. The force exerted by the robot-assisted tool 22 on the workpiece 24 is then increased to the desired force setpoint ( Fig. 12D), as further described herein. Machining of the workpiece 24 continues with the FCD 10 moving along its compliance stroke and / or adjusting the force behind the piston 16 to maintain the applied force at the force setpoint. As one example, the applied force may be adjusted to account for changes in a component of the tool weight vector as the robotic tool 22 changes its orientation in the inertial reference frame.

[0041] When the programmed machining of the workpiece 24 is completed, the robot-based controller moves the robot arm away from the workpiece 24. The FCD 10 moves the robot-based tool 22 in a compliance movement ( Fig. 12B) to maintain contact with the workpiece 24 with the desired contact force ( Fig. 12D). If contact is lost ( Fig. 12A), the robot-assisted tool 22 hits the mechanical stop (or the programmed position tolerance) ( Fig. 12C), which is registered as a contact loss event ( Fig. 12A). The robot-assisted tool 22 then returns to the search position ( Fig. 12B, Fig. 12C).

[0042] According to a second aspect of the present disclosure, a transition search method is used to detect contact between a robotic tool 22 and a workpiece 24. The FCD 10 is configured as shown in Fig. 13. The robotic tool 22 is moved to an initial position at or near the outermost extent of the compliance travel of the FCD 10 (farthest from the workpiece 24). As shown, the software can be programmed to ignore the position tolerance for this initial position. In the transition search process, the robot arm can be positioned at an estimated distance from the workpiece 24 that is less than the compliance travel of the FCD 10.

[0043] The FCD 10 then slowly advances the robot-assisted tool 22 toward the workpiece 24 while monitoring the force exerted by the robot-assisted tool 22. The contact between the robot-assisted tool 22 and the workpiece 24 is determined by detecting a contact force F d that exceeds a predetermined threshold is detected. In response to the detection of the contact, the movement of the FCD 10 is stopped (or alternatively controlled to bring the FCD 10 into the desired position, as described in Fig. 13). Loss of contact as the robot arm moves away from the workpiece 24 is detected as in the steady-state search method described above, ie, by the position of the robotic tool 22 being within the position tolerance of the FCD 10 (ie, within the threshold of the mechanical stop or touching it).

[0044] Fig.14A shows the actual contact, and Fig. 14B-D show the speed, position, and force of the robotic tool 22 for the transition search procedure. Initially, the robotic tool 22 moves toward the workpiece ( Fig. 14B, Fig. 14C). At and after the point of contact ( Fig. 14A), the measured force with which the robot-supported tool 22 is pressed against the workpiece 22 increases ( Fig. 14D). The increase of this force above a threshold signals contact ( Fig. 14d). The detection of the contact stops the forward movement of the robot-assisted tool 22 ( Fig. 14B, Fig. 14C). The force exerted by the robot-assisted tool 22 on the workpiece 24 is then increased to the desired contact force, as described in more detail here ( Fig.14D). Machining of the workpiece 24 continues, with the FCD 10 adjusting the force behind the piston 16 to maintain the applied force at the force setpoint. Following machining, when the robot arm moves the robot-supported tool 22 away from the workpiece 22 (or vice versa), loss of contact detection occurs as previously described with reference to the steady-state search method.

[0045] Once contact between the robotic tool 22 and a workpiece 24 is detected—either through the steady-state search or transition techniques described above—the FCD 10 transitions from position control to force control. To avoid force spikes that could damage the workpiece 24, this transition is ideally smooth. Smooth switching is known in the art as a technique used when switching a proportional-integral-derivative (PID) controller between control modes—typically between manual and automatic control modes. The desired result of smooth switching is to avoid transient changes in the controller output when changing modes.As is well known in the art, a PID controller continuously calculates an error value as the difference between a desired setpoint and a measured process variable (PV). Proportional, integral, and derivative control loops then each generate contributions to a controlled variable that alters a process or property to achieve a change in the process variable toward the setpoint. Bumpless switching ensures that changes in the controlled variable are gradual, resulting in only a gentle change in the process variable as a result of the change in control mode.

[0046] According to aspects of the present disclosure, smooth control switching is used to switch the FCD 10 from position control mode to force control mode. Prior to contact with a workpiece 24, the FCD 10 operates in position control mode. The inputs of the PID controller are position, velocity, and force, and the outputs control the pneumatic pressure in the front and rear chambers of a pneumatic cylinder 14 to control the position of the piston 16 (and thus the carriage 20 connected to the robotic tool 22, in the example of Fig.1). After contact is detected between the robotic tool 22 and the workpiece 24, the FCD 10 switches to a force control mode. The inputs to the controller are still position, velocity, and force, but the outputs control the pneumatic pressure in the cylinder 14 to control the force exerted by the robotic tool 22 on the workpiece 24. By employing a smooth transition method when switching between the position and force control modes, the pressure changes in the chambers of the pneumatic cylinder 14 are smooth and do not produce large transients in the force exerted on the workpiece 24. In particular, the contact force does not exceed the force applied to the workpiece 24, unlike the graph in Fig. 4 does not specify the force setpoint, ie the desired continuous force exerted between the robot-supported tool and the workpiece.

[0047] Fig.Figure 15 shows a basic block diagram of a high-level FCD 10 PID controller. The blocks surrounded by dashed boxes are the variables or functions that experience a change in engineering units / quantities upon transition to control mode. Immediate changes to these parameters would directly result in an immediate change in the output. To prevent the output from changing immediately upon switching control mode, the transition procedure is controlled for each of these identified variables / functions.

[0048] Table 1 lists the control terms that change immediately from one iteration of control in a first mode (i.e., position control) to the next iteration of control in a different mode (i.e., force control). No transition mechanism is required for these terms. In Table 1, k = 1 defines the first iteration of the respective control mode. Table 1: Change of control terms when changing the control mode Control term Position control value @k=1 Force control value @k=1 theoretical net force g * total mass g * total mass + force setpoint Setpoint user-specific position setpoint user-specified force setpoint PV Position feedback (x) force feedback (f) PV rate Velocity feedback (dx / dt) Force rate feedback (df / dt) Kp Position Kp Reinforcement Force Kp Reinforcement Ki Position Ki reinforcement Power Ki Enhancement Customer Position Kd Reinforcement Force Kd Reinforcement

[0049] For a smooth transition between control modes, the units and relative quantities listed in Table 1 must transition between their values ​​in position mode and force mode in such a way that no jump occurs in the control output. The setpoint filter must be reset; however, a simple reset to zero would cause the filter to change more drastically than desired under certain conditions (i.e., position and force combinations). Accordingly, the control mode transition is more detailed.

[0050] In particular, both the setpoint and the process variable (PV) change from displacement to force units. The integral term of the PID controller must be reset to account for these scaling differences between the units. Resetting the integral term causes a discontinuous change in the output value, so a simple reset alone is not sufficient to achieve a bumpless transition. Accordingly, resetting the integral term decays over many iterations.

[0051] The setpoint filter is a low-pass filter that smooths setpoint commands. With high-frequency changes or large amplitude changes, the PID controller is more susceptible to instability. The setpoint filter therefore limits the setpoint changes for each iteration of the controller to a level that ensures control stability. When changing the PID controller operating mode, the setpoint filter value is phased out, meaning the filter coefficients are updated only slightly, to values ​​that allow the filter to maintain the current value as if it were already in steady state. In one aspect, a linear low-pass filter is used so that the filter coefficients can be easily calculated and initialized to values ​​that reflect a steady-state filter value.

[0052] To transition the PID controller from position control mode to force control mode, the control gains Kp, Ki, and Kd are changed from position gains to force gains. The setpoint is changed to the force setpoint, which has different physical units. The internal state of the setpoint filter is reset, and the filter is iteratively adjusted to the starting value of the new setpoint.

[0053] Another point to consider when transitioning the PID controller from position to force control mode is the integral term. An immediate change in the integral term from the position mode value to a force mode value can cause a sudden change in the pressure of the pneumatic cylinder 14 and a potentially damaging discontinuity in the operation of the FCD 10. Accordingly, when changing the PID controller mode, the current value of the integral term is copied to a memory location associated with a previous value of the integral term and used there for multiple iterations of the PID controller. The current value of the integral term decays to zero over multiple iterations in one aspect by simple exponential decay: 1(1−scaling factor)i∗previous integral term where the discrete control iteration i starts at 1 (if it starts at 0, (i+1) is used in equation (6) instead of i to avoid division by zero problems).

[0054] The transition of the integral term is active for i < X, where X is a time constant (expressed as the number of control iterations) sufficient to complete the bumpless transition. The preceding integral term is the integral term of (Integral (Error) * K i ) from the start of the mode transition.

[0055] For i <X ist der Integralterm der PID-Steuerung Integral term=1((scaling factor∗i)i∗previous integral term+current integral term∗Ki

[0056] And for i ≥ X, the integral term of the PID control is Integral term=current integral term∗Ki

[0057] Fig.Figure 16 is a graph of the force between the robotic tool 22 and a workpiece 24, according to aspects of the present disclosure. Initially, the robotic tool 22 is held at a position along the compliance travel of the FCD 10, away from a mechanical stop, and generates a net force of zero. The force increases upon contact of the robotic tool 22 with the workpiece 24. However, the force at initial contact, or contact force, is well below the force setpoint. By smoothly switching the PID controller from position control mode to force control mode and gradually ramping the applied force up to the force setpoint, the applied force increases smoothly and without sharp transients. Importantly, the initial contact force never exceeds the force setpoint.

[0058] Fig.17 shows the steps of a method 100 for controlling a robotic force control device 10 comprising a robotic tool 22. The robotic force control device 10 has a carriage 20 capable of moving along a compliance stroke limited at least at one end by a mechanical stop. In a position control mode, the robotic tool 22 is positioned along a compliance stroke of the force control device 10 away from a mechanical stop in the force control device 10 (block 102). The robotic tool 22 and / or a workpiece are moved toward each other (block 104). Contact between the robotic tool 22 and the workpiece is detected (block 106). A smooth transition from the position control mode to the force control mode occurs (block 108).In force control mode, a force exerted by the robot-assisted tool 22 on the workpiece is increased to a force setpoint (block 110).

[0059] In one aspect, the force applied to a workpiece is increased via a linear ramp to the final force value desired by the user. This linear ramp is implemented as a discrete ramp rate filter with the force controller setpoint. The time-domain filter implementation is achieved by the simple update equations: ΔSPk=SPdesired−SPk−1 SPk=SPk−1+sign(ΔSPk)⋅min(setpoint ratezul,ΔSPk), where

[0060] SP erwünscht the desired setpoint is, SP k-1 is the setpoint from the previous iteration (or is initialized to the current value of the feedback for this controlled term when the controller is started), ΔSPk is the filtered setpoint, which is then transferred to the respective controller, and Setpoint rate defines the maximum force setpoint change per iteration of the control; and thus takes the frequency of the control into account.

[0061] A value of force units per iteration is derived from defining a ramp in force units per second and a known control frequency. This is used in the force control, the setpoint filter. In other aspects, the ramp from the initial force to the desired force can follow a user-defined trajectory by replacing Equation (10) with a desired transition function.

[0062] A pneumatic cylinder piston is often modeled as a mass between two springs due to the spring-like nature of compressed air. The lower the pressures on either side of the piston, the lower the mechanical stiffness of the system. Typically, an FCD includes low-friction components, so system stiffness is very low at low pressures. This is ideal for an FCD operating in force control mode. However, operating the FCD in position control mode can result in oscillating behavior. In particular, horizontal orientation (with respect to gravity) represents a very low-stiffness system because the pressures within the cylinder are very low in position control mode. This can result in the system being unable to perform stable and repeatable position control. In order for the searchPosition orIn order for the search position to be held effectively, the position control must have sufficient rigidity to maintain the desired position without undesirable oscillation behavior.

[0063] To increase the stiffness of the pneumatic system, the device increases the cylinder's backpressure. There are many known methods for increasing the backpressure of a device, but to be effective in the FCD, it is also necessary to return to a minimally stiff system when the backpressure is removed. Therefore, the FCD should have a way to effectively change the system's stiffness dynamically.

[0064] Two aspects that enable this are described here. These are presented to thoroughly explain the concept to those skilled in the art and are not intended to be the only means of performing such a function.

[0065] Fig.Figure 18 shows such an aspect, in which the chambers of cylinder 14 are pressurized by independent pressure control devices 30, 32. One pressure control device 30 regulates the backpressure—which is set to atmospheric pressure in force control mode, but increased to a higher pressure value in position control mode. Another pressure control device 32 regulates the pressure in the other chamber depending on the desired system stiffness and to control the pressure exerted on the workpiece.

[0066] Fig.Figure 19 shows another aspect in which a single pressure control device 34 directs pneumatic fluid through a directional control valve 36 to one of the two cylinder chambers. The cylinder chamber not connected to the pressure control device 34 is connected to the outlet. The outlet flow path is configured to have a high-resistance path 38 and a low-resistance path 40. A three-port, two-state directional control valve 42 is used to switch between the two flow paths 38, 40. The low-resistance path 40 is directly connected to the outlet and offers very little resistance to the outlet flow. Such a flow path 40 provides minimal backpressure in the device and is therefore the flow path used in force control mode.The high resistance flow path 38 consists of a flow restrictor 44 that limits the discharge amount, thereby increasing the back pressure in the device and keeping the FCD 10 more stable in position control mode.

[0067] When contact with the workpiece is detected and the transition from position control mode to force control mode is triggered, the system's stiffness is immediately changed to enter the minimum stiffness state. Upon loss of contact, the FCD 10 returns to position control mode, which triggers the higher stiffness configuration. An FCD 10 in a stiff configuration is a system where the backpressure in the cylinder is sufficient to prevent oscillation in position control mode. An FCD 10 in a minimal stiffness configuration is a configuration where the cylinder backpressure is low, enabling fine force control in force control mode but may allow oscillation in position control mode. In various aspects of the present disclosure, the FCD 10 may assume a stiff configuration in position control mode and transition to a minimal stiff configuration in force control mode.

[0068] Aspects of the present disclosure offer numerous advantages over the prior art. By adding one or more force overshoot mitigation air passages—either connecting chambers through piston 16 or venting air from a chamber through cylinder 14 or the air line(s)—the force overshoot upon initial contact of a material removal tool 22 with a workpiece 24 is drastically reduced. This can lead to more consistent results in the finishing of workpieces and avoid harmful effects that can result from excessive contact force.

[0069] In general, all terms used herein should be interpreted according to their ordinary meaning in the relevant technical field, unless another meaning is clearly stated and / or is clear from the context in which it is used. All references to an element, device, component, means, step, etc. should be construed to refer to at least one instance of the element, device, component, means, step, etc., unless expressly stated otherwise. The steps of the methods disclosed herein do not need to be performed in the exact order unless a step is expressly described as following or preceding another step and / or it is implied that a step must follow or precede another step.Any feature of any aspect disclosed herein may be applied to any other aspect, where appropriate. Likewise, any benefit of any aspect may apply to any other aspect, and vice versa. Other objectives, features, and benefits of the accompanying aspects will be apparent from the description. As used herein, the term "configured to" means arranged, organized, adapted, or arranged to operate in a particular manner; the term is synonymous with "designed to."

Claims

[1] A method for controlling a robot-assisted force control device (10) comprising a robot-assisted tool (22), comprising: in a position control mode, positioning the robot-assisted tool (22) along a compliance stroke of the force control device (10) away from a mechanical stop in the force control device (10); in the position control mode, moving the robot-assisted tool (22) and / or a workpiece (24) towards each other; in the position control mode, detecting contact between the robot-assisted tool (22) and the workpiece (24), wherein a force exerted by the robot-assisted tool (22) on the workpiece (24) upon contact is less than a force setpoint; Performing a smooth transition from the position control mode to a force control mode; and in the force control mode, increasing the force exerted by the robot-assisted tool (22) on the workpiece (24) to the force setpoint. [2] The method of claim 1, wherein the force control device (10) comprising the robotic tool (22) is mounted on a robot, and wherein the robotic tool (22) is moved toward the workpiece (24) to make contact. [3] The method of claim 1, wherein the force control device (10) comprising the robot-assisted tool (22) is stationary, and wherein the workpiece (24) is moved toward the robot-assisted tool (22) to make contact. [4] A method according to any one of the preceding claims, wherein detecting contact between the robot-assisted tool (22) and the workpiece (24) comprises searching for a steady state, comprising: Positioning the robot-assisted tool (22) at an initial search position along a compliance movement stroke of the force control device (10); and Moving the robot-assisted tool (22) and the workpiece (24) towards each other; wherein detecting contact between the robot-assisted tool (22) and the workpiece (24) comprises detecting a change in the position of the robot-assisted tool (22) along the compliance movement stroke of the force control device (10). [5] A method according to claim 3 or 4, wherein the initial search position is within one half of the compliance movement stroke of the force control device (10) in the direction of the workpiece (24). [6] A method according to any one of the preceding claims, wherein detecting contact between the robotic tool (22) and the workpiece (24) comprises a search for a steady state, comprising: Positioning the robot-assisted tool (22) at an initial search position along a compliance movement stroke of the force control device (10); and Moving the robot-assisted tool (22) and the workpiece (24) towards each other; wherein detecting contact between the robotic tool (22) and the workpiece (24) comprises detecting a force with which the robotic tool (22) contacts the workpiece (24) that exceeds a predetermined threshold. [7] A method according to any one of the preceding claims, wherein detecting contact between the robotic tool (22) and the workpiece (24) comprises a transition search comprising: Positioning the robot-assisted tool (22) at an initial position along a compliance stroke of the force control device (10); and Moving the robot-assisted tool (22) along the compliance stroke of the force control device (10) towards the workpiece (24); wherein detecting contact between the robotic tool (22) and the workpiece (24) comprises detecting a force with which the robotic tool (22) contacts the workpiece (24) that exceeds a predetermined threshold. [8] The method of claim 7, wherein the initial position is located near a mechanical stop in the compensation movement stroke of the force control device (10) opposite to the workpiece (24). [9] A method according to any one of the preceding claims, further comprising detecting a loss of contact between the robotic tool (22) and the workpiece (24) by detecting a position of the robotic tool (22) along the compliance stroke of the force control device (10) that is a predetermined distance from a mechanical stop closest to the workpiece (24). [10] A method according to any one of the preceding claims, wherein the force control device (10) comprises a proportional, integral, derivative (PID) controller, and wherein performing the bumpless transition from the position control mode to the force control mode comprises: Changing a position command to a force command with other physical units; Resetting an internal state of a low-pass setpoint filter; and iteratively updating coefficients of the setpoint filter so that an output of the setpoint filter reaches the force setpoint over a predetermined number of control iterations. [11] The method of claim 10, wherein performing a smooth transition from the position control mode to the force control mode comprises: Changing an integral gain from a position integral gain to a force integral gain; for a transition period comprising a predetermined number of control iterations after switching from the position control mode to the force control mode, ramping a value of an integral term of the PID controller from a value calculated using the position integral gain to a value calculated using the force integral gain. [12] The method of claim 11, wherein ramping a value of the integral term of the PID controller from a value calculated using the position integral gain to a value calculated using the force integral gain comprises: when switching from the position control mode to the force control mode, copying a current integral term into a memory location assigned to a previous integral term; and Reset the current integral term; during the transition period, at each control iteration, Decay of the value of the previous integral term; and Setting an integral term of the PID controller to the decayed previous integral term plus the current integral term times the force integral gain; and after the transition period, at each control iteration, Set the integral term to the current integral term times the force integral gain. [13] The method of claim 12, wherein the decay of the value of the previous integral term comprises an exponential decay of the value of the previous integral term. [14] Method according to claim 12 or 13, wherein during the transition period, Integral term=1(scaling factor∗i)i∗previous integral term+current integral term∗Ki; and after the transition period, Integral term=current integral term∗Ki, where K i is the force integral gain factor; i is the discrete control iteration, starting with 1 at the control mode change; and the scaling factor is a user-defined scaling factor. [15] A method according to any one of the preceding claims, wherein performing the bumpless transition from the position control mode to the force control mode further comprises switching from a stiff configuration to a minimally stiff configuration. [16] A method according to any one of the preceding claims, wherein increasing the force exerted by the robot-assisted tool (22) on the workpiece (24) to a force setpoint comprises increasing the force linearly. [17] The method of any of claims 1-15, wherein increasing the force exerted by the robotic tool (22) on the workpiece (24) to a force setpoint comprises increasing the force along a user-defined trajectory. [18] A force control device having a compliance movement and configured to hold a robot-supported tool (22) and apply the robot-supported tool (22) with a desired force against a workpiece (24), comprising: a controller (26) operating in a position control mode to to position the robot-assisted tool (22) along a compliance stroke of the force control device (10) away from a mechanical stop in the force control device (10); and to detect contact between the robot-assisted tool (22) and a workpiece (24) when the robot-assisted tool (22) and / or the workpiece (24) move towards each other, wherein a force exerted by the robot-assisted tool (22) on the workpiece (24) upon contact is less than a force setpoint; wherein the controller (26) further operates in a force control mode to: to increase a force exerted by the robot-assisted tool (22) on the workpiece (24) to the force setpoint; wherein the controller (26) is configured to perform a smooth transition from the position control mode to the force control mode. [19] A force control device according to claim 18, further comprising: a position sensor configured to detect a position of the robot-assisted tool (22) along a compliance movement stroke of the force control device (10); and a force sensor configured to detect a force with which the robot-assisted tool (22) contacts the workpiece (24). [20] A force control device according to claim 18 or 19, wherein the controller (26) is configured to detect contact between the robot-assisted tool (22) and the workpiece (24) by a steady-state search, comprising: Positioning the robot-assisted tool (22) at an initial search position along a compliance movement stroke of the force control device (10); and the controller (26) is configured to detect contact between the robot-assisted tool (22) and the workpiece (24) as the robot-assisted tool (22) and the workpiece (24) move toward each other by detecting a change in position of the robot-assisted tool (22) along the compliance movement stroke of the force control device (10). [21] A force control device according to claim 20, wherein the initial search position is within one half of the compliance movement stroke of the force control device (10) in the direction of the workpiece (24). [22] A force control device according to any one of claims 18-21, wherein the controller (26) is configured to detect contact between the robot-assisted tool (22) and the workpiece (24) by a steady-state search, comprising: Positioning the robot-assisted tool (22) at an initial search position along a compliance movement stroke of the force control device (10); and the controller (26) is configured to detect contact between the robot-assisted tool (22) and the workpiece (24) as the robot-assisted tool (22) and the workpiece (24) move toward each other by detecting a force with which the robot-assisted tool (22) contacts the workpiece (24) that exceeds a predetermined threshold. [23] Force control device according to one of claims 18-22, wherein the controller (26) is configured to detect the contact between the robot-assisted tool (22) and the workpiece (24) by searching for a transition, comprising: Positioning the robot-assisted tool (22) at an initial position along a compliance stroke of the force control device (10); Moving the robot-assisted tool (22) along the compliance stroke of the force control device (10) in the direction of the workpiece (24); and Detecting contact between the robot-assisted tool (22) and the workpiece (24) by detecting a force with which the robot-assisted tool (22) contacts the workpiece (24) that exceeds a predetermined threshold. [24] A force control device according to claim 23, wherein the initial position is located near a mechanical stop in the compensation movement stroke of the force control device (10) opposite to the workpiece (24). [25] The force control device of any one of claims 18-24, wherein the controller (26) is further configured to detect a loss of contact between the robotic tool (22) and the workpiece (24) by detecting a position of the robotic tool (22) along the compliance stroke of the force control device (10) that is a predetermined distance from a mechanical stop closest to the workpiece (24). [26] A force control device according to any one of claims 18-25, wherein the controller (26) comprises a proportional, integral, derivative (PID) controller, and wherein the PID controller is configured to perform a bumpless transition from the position control mode to the force control mode by: a position command is changed into a force command with other physical units; an internal state of a low-pass setpoint filter is reset; and Coefficients of the setpoint filter are iteratively updated so that an output of the setpoint filter reaches the force setpoint over a predetermined number of control iterations. [27] A force control device according to claim 26, wherein the PID controller is further configured to perform a bumpless transition from the position control mode to the force control mode by: an integral gain is changed from a position integral gain to a force integral gain; for a transition period comprising a predetermined number of control iterations after switching from the position control mode to the force control mode, a value of an integral term of the PID controller is ramped from a value calculated using the position integral gain to a value calculated using the force integral gain. [28] A force control device according to claim 27, wherein ramping a value of an integral term of the PID controller from a value calculated using the position integral gain to a value calculated using the force integral gain comprises: when switching from the position control mode to the force control mode, Copying a current integral term into a memory location assigned to a previous integral term; and Reset the current integral term; during the transition period, at each control iteration, Decay of the value of the previous integral term; and Setting an integral term of the PID controller to the decayed previous integral term plus the current integral term times the force integral gain; and after the transition period, at each control iteration, Set the integral term to the current integral term times the force integral gain. [29] A force control device according to claim 28, wherein the decay of the value of the previous integral term comprises an exponential decay of the value of the previous integral term. [30] Force control device according to claim 28 or 29, wherein during the transition period, Integral term=1((scaling factor∗i)i∗previous integral term+current integral term∗Ki; and after the transition period, Integral term=current integral term∗Ki, where K i is the force integral gain factor; i is the discrete control iteration, starting with 1 at the control mode change; and the scaling factor is a user-defined scaling factor. [31] A force control device according to any one of claims 18-30, wherein the controller (26) is further configured to assume a stiff configuration in the position control mode and a minimally stiff configuration in the force control mode. [32] Force control device according to one of claims 18-31, wherein the controller (26) is further capable of linearly increasing a force exerted by the robot-assisted tool (22) on the workpiece (24) to a force setpoint upon transition from the position control mode to the force control mode. [33] The force control device of any of claims 18-31, wherein the controller (26) is further capable of increasing a force exerted by the robot-assisted tool (22) on the workpiece (24) to a force setpoint by increasing the force along a user-defined trajectory when transitioning from the position control mode to the force control mode.

Images