GB2536227A

GB2536227A - Pipettor Autoteaching

Info

Publication number: GB2536227A
Application number: GB1503940.7A
Authority: GB
Inventors: Rech Thomas; Leibfried Thomas; Tahedl Harald
Original assignee: Stratec Biomedical AG
Current assignee: Stratec Biomedical AG
Priority date: 2015-03-09
Filing date: 2015-03-09
Publication date: 2016-09-14
Also published as: GB201604016D0; GB2540646A; GB201503940D0

Abstract

A method for determining reference points in automated pipetting system comprising the steps of defining a coordinate system in the automated pipetting system having two horizontal axes and one vertical axis, generating a map of reference points and their nominal position in the coordinate system, detecting automatically the actual position of the reference points by moving sensors, comparing the detected actual position with the nominal position of the reference points, determining deviations between the actual and nominal position of the reference points and adjusting the position of the reference points within the coordinate system corresponding to detected position deviations. The sensors maybe capacity sensors or force feedback sensors. Ideally, four nominal reference points are defined in the co-ordinate system with a 5mm maximum catch range defined around each nominal reference point. The method allows of the position teaching or calibration of an automated or robotic pipetting system.

Description

Pipettor Autoteaching

Field of the Invention

[0001] The field of the invention relates to a method for determining reference points in au-tomated pipetting system.

Background of the invention

[0002] Automated analyser systems for use in clinical diagnostics and life sciences are produced by a number of companies. For example, the Stratec Biomedical AG, Birkenfeld, Ger- produces a number of devices for specimen handling and detection for use in automat-ed analyser systems and other laboratory instrumentation.

[0003] Processes in the use of laboratory equipment such as automated analyser systems for clinical diagnostics are being increasingly automated. High-throughput technologies require management of resources in respect of components of the processed tests or samples as well as in respect of information. Some analyses may require a number of different components, such as reagents, for any series of tests or samples. The user of the automated analyser system needs to manage the large number of tests or samples needs with the help of information gathering and processing.

[0004] The location of the positions has to be trained at least on a first set-up of a pipetting system in order to precisely access positions in a robotic pipetting system. Usually the process has to be repeated after any intervention within the system that can lead to a shift of positions.

[0005] The current standard procedure is to access a number of reference positions using the robotic system via manual movement commands. The operator aligns the robot and the reference positions visually. The location of the aligned robot is then read out and transferred in a robot coordinate space. This process is called "teaching" of the robotic positions.

[0006] It is possible to interpolate coordinates in an area framed by the reference points by determining multiple reference point coordinates (usually two or three) in robot coordinates.

[0007] Correct teaching procedures use three corner points to interpolate array positions in a parallelogram shape in particular for rectangular arrays of positions. The coordinate grid within the parallelogram is limited to regular spacing in the two parallelogram axes.

[0008] The visual alignment of the robot and reference point has a number of fairly obvious disadvantages: It is very tedious, since usually a significant number of individual reference points have to be determined.

Errors in the determined robot coordinates of the reference point locations cannot be quantified due to subjectivity of the visual alignment process.

The tediousness of the process makes multiple repeat coordinate determinations to average out errors intolerable.

The number of reference points has to be reduced to the bare minimum making redundancy to reduce errors impossible. In particular the interpolation of parallelogram ar- rays with only three reference points accumulates all teaching errors in the fourth cor-ner, which is not directly determined Object of the Invention [0009] It is an object of the present invention to provide a method for a reliable determination of reference points in in automated pipetting system.

Summary of the Invention

[0010] The present disclosure relates to a method for determining reference points in automated pipetting system, comprising the steps of a. defining a coordinate system in the automated pipetting system having two horizontal axes and one vertical axis; b. generating a map of reference points and their nominal position in the coordinate system; c. detecting automatically the actual position of the reference points by moving sensors; d. comparing the detected actual position with the nominal position of the reference points; e. determining deviations between the actual and nominal position of the reference points; and f. adjusting the position of the reference points within the coordinate system corresponding to detected position deviations.

[0011] The sensors may be capacity sensors or force feedback sensors and at least four nomi-nal reference points in the coordinate system can be defined and detected.

[0012] A catch range may be defined around each nominal reference point, wherein the catch range may be maximal 5 mm around a nominal position of a reference point.

[0013] It is further envisaged that steps a) to 0 can be repeated at least once to ensure con-sistency and to identify errors.

[0014] It is further intended that the three axes are scanned consecutively.

[0015] The target may be grounded when using capacity sensors.

[0016] The sensors of a robotic system of the automated p petting system may be used for detection.

[0017] It is further envisaged that the actual positions of reference points may be interpolated.

[0018] Intersecting lines or planes may also be detected with a method according to the instant disclosure.

[0019] The adjustment of the position of the reference points within the coordinate system may depend on predefined thresholds for maximal allowable deviations.

[0020] The detected actual position in the vertical axis may be fitted in a plane with minimal deviations or alternatively a non-flat surface may be generated going through actual positions in the vertical axis.

[0021] The nominal positions may be generated as a CAD output or by physically measuring the nominal positions.

Summary of the Figures

[0022] The invention will be described on the basis of figures. It will be understood that the embodiments and aspects of the invention described are only examples and do not limit the protective scope of the claims in any way. The invention is defined by the claims and their equivalents. It will be understood that features of one aspect or embodiment of the invention can be combined with a feature of a different aspect or aspects and/or embodiments of the invention. It shows: [0023] Figure 1 Use of unbiased mapping of nominal positions within a tetragon in CAD space to a tetragon in robotic coordinates [0024] Figure 2 Catch range around starting vale for each corner [0025] Figure 3 Probe straightness determination [0026] Figure 4 Area between four Z positions Detailed Description of the Invention and the Figures [0027] Robotic systems for liquid pipetting are equipped with a number of sensors. These sensors can be utilised to detect the reference point locations with no added cost for the robotic system if the reference points are suitably realised.

[0028] Two sensoric systems are particularly useful for this purpose: a. The capacity sensor for detecting the surface of the liquid for pipetting is able to detect any object which changes the value of the capacitor build by pipettor needle and grounded instrument housing.

b. The drive mechanism of the robot uses a force feedback loop for motion control. This can be used to detect collisions with a mechanically solid object or surface.

[0029] Both methods allow automatic determination of the location of reference points. This takes the subjectivity and tediousness out of the alignment process and allows for a. a repeated determination of the position to ensure consistency and quantify position errors.

b. the determination of as many reference points as necessary to achieve the desired precision is further possible.

[0030] Using sensors for determining reference points, in particular for rectangular areas of positions, it becomes feasible to determine the fourth corner and use unbiased mapping of nominal positions within a tetragon in CAD space to a tetragon in robotic coordinates. This means that the assumption that the robotic coordinates form a parallelogram is abrogated and therefore the fourth corner no longer accumulates all teaching errors. Further the use of direct mapping mathematics allows to map an arbitrary pattern of nominal (e.g. CAD defined) positions to robot access positions (comp figure 1).

[0031] For capacitive position detection a good way to define a target position is to use three intersecting lines or planes or a combination thereof in automatic position detection with a robotic probe. The intersection marks the position of the reference point. Using two edges in X and Y and a plane in Z are well suited for a pipettor. This means that the corner of a block or a blade can define a reference position.

[0032] Such a reference feature has to change the capacity of the probe significantly when being touched. Ideally it should be grounded if a capacity sensor is used.

[0033] The robot scans the three motion axes consecutively with activated capacity detection. When the target is touched a jump in capacity triggers the stop of the motion. Thereby the position of the target edge in each axis is found.

[0034] The size required for the feature depends on the desired catch radius around the marked position from within which a successful search is expected. The same argument defines the required free space around the marked position. Since corners are used they can be oriented in different ways in the XY-plane.

[0035] A catch range (comp. figure 2, circles) of about 5mm from a starting value appears a reasonable compromise between tombstone size and deviation allowance that might result from various actions that could trigger re-teaching. (like probe exchange, module exchange etc.) [0036] This means that the edges defining the corner must be longer than 5mm, at least 6mm, better 8mm.

[0037] When capacitive detection is used to determine the location of hard targets great care must be taken in the software that the probes are not destroyed in the process. In particular, when moving in X and Y sideways against the targets the movement range must be tightly limited about an expected position, otherwise the probes can be bent or even broken.

[0038] The targets should ideally be well grounded to obtain a perfect signal jump. If ground-ing cannot be assured a sufficient metal mass is required.

[0039] The reference position search can be repeated a few times for every axis to ensure reproducibility. The consistency of the identified positions can be checked to eliminate false positive detections or delayed detection.

[0040] By using a horizontal blade to form the target edges with free height of at least a probe length below, it becomes possible to touch the edge with the tip and the shoulder of the probe sequentially. Thereby the tilt of the probe in both horizontal axes can be determined. The tilt is simply the difference between detected positions P1 and P2 (comp figure 3).

[0041] When multiple pipetting probes are used on a single robotic gantry their relative positions cannot be independently moved in the common axis. In such cases it is advantageous not to use the positions determined by a single one of the joined probes. With the reproducibility of the capacitive detection system it is useful to determine the robot coordinate of one target with each of the joined probes and then use the median of that position as reference. For fur-ther positions the offset between the median and the position determined by any one of the probes can be used to correct further positions determined with any one of the probes.

[0042] Both, probe straightness determination and multi probe alignment correction should be done on a single target before any further targets are teached.

[0043] Obviously for both, straightness and multi probe alignment, it is reasonable to define thresholds for allowable deviations. If the thresholds are exceeded further teaching or even usage of the robotic system is not sensible and suitable software locks can be used to prevent further usage.

[0044] For the second usable sensor system -using force feedback sensors -the target posi-tions must be shaped as elevated platforms. The search is done by driving the vertical axis slowly downwards under force feedback. When the pipetting probe hits the elevated block the force feedback stops the motion. When the probe is not above the elevated platform the drive stops at a predefined height without hitting an obstacle. By scanning for the obstacle along each one of the horizontal axes consecutively the location of the edges of the obstacle can be found. The height of the obstacle is defined by the vertical position at which the force feed-back stops the motion.

[0045] Another time effective approach requires the starting position to be above the platform. The vertical move hits the platform and yields the platform height. A defined distance, larger than the platform size, shifts the horizontal position. This is expected to be a low posi-tion where at platform height minus free height no collision occurs. Now half the distance between the high and low points along the selected axis is moved back and the vertical move is repeated. This process is repeated until the edge of the platform is defined with the resolution of the horizontal axis.

[0046] In order to minimize errors caused by the probe slipping sideways off the edge of the target platform, the platform is not just a corner, as is sufficient for capacitive scanning. The platform is shaped as square block and both edges of the platform along each axis are scanned. The target position in this case is not the intersection of platform edges, but the cen-ter of the platform.

[0047] For platform size and free space around the target the same considerations as above apply. The platform and the robotic probe need to be able to withstand the mechanical load exerted before the force feedback stops the motion. For sensitive probes a protective cover can be useful.

[0048] In order to avoid determination of robotic coordinates for each and every point that the robot will access, it is common practice to determine the corner points of rectangular areas within which the access positions are mechanically stable relative to each another ( e.g. within a mechanical subassembly). The access positions are then interpolated within the area derived from the determined corner points.

[0049] STRATEC® systems currently use a teacher software with three corner points that allows for regular parallelogram grids with offsets from the three corner points. This method has certain disadvantages: a. All teaching errors in the three corner points add up in the extrapolated fourth corner.

b. Irregular patterns of access positions require multiple interpolation areas even when they are mechanically stable relative to one another.

c. Trapezoidal mechanical distortions are not accounted for.

d. The nominal corner point geometry has to be a parallelogram [0050] These disadvantages can be overcome by using mathematical mapping of a tetragon in nominal coordinates to a tetragon in robotic coordinates The mapping process requires a map of nominal positions as input. This can be generated by CAD output or by physically measuring the relevant positions on a unit. The robotic coordinates of four (or more) corner points are determined by a teaching process either automatic or manually [0051] The equations for mapping a coordinate (X,Y) within an arbitrary tetragon to a coor-dinate (X',Y' within another arbitrary tetragon are online available. They are an I ais

I

fang i r1 I' I i.12_t I cw [0001] The nine mapping coefficients an... a3s are determined by solving the eight equations that the determination of the four corner positions in robotic coordinates provide and the standardisation a33 = I. With the mapping coefficients known any nominal coordinate can be mapped to a robotic coordinate.

[0002] With four determined values for the Z coordinate the assumption that the Z coordinates form a tilted plane is not necessarily valid any more. In fact in most cases the four measured Z positions will not lie in a plane (comp figure 4).

[0003] There are two straightforward approaches to solve the issue.

a. Fit a plane with minimal deviations to all four corners.

b. Replace the plane by a non-flat surface that goes through all four corner points. [0052] The simplest way to achieve that all Z positions lie within a plane is to calculate the Z level from the Z level of all four corners weighed by their distance to the position to be calcu-lated. With four known corner points (Xi, Yi, Z1) to (X4, Y4, Z4) the Z value for any point (X, Y) is ( X. Y1 Iii. ,:' , [0053] Advantages of the instant disclosure concerning a method for automatic position determination using force feedback can be summarized as follows: - The process has a definable reproducability - The process is fully automatic, therefore o it can determine as many points as useful o it can be repeated to confirm the reproducability No extra cost, as existing sensors are used [0054] Advantages of the instant disclosure concerning a method for automatic position determination using capacitive position determination can be summarized as follows: The process has a definable reproducability The process is fully automatic, therefore o it can determine as many points as useful o it can be repeated to confirm the reproducability - No extra cost, as existing sensors are used - Position finding can be done by direct scanning to the detectable target.

- Mechanical load on the robot probe is negligible.

[0055] Advantages of the instant disclosure concerning a method for automatic position determination using tetragon mapping can be summarized as follows: Reduced error since 4 instead of 3 corners are actually determined and used for map-ping.

Direct calculation of robot coordinates from a fix map of nominal positions. All types of linear distortions of the tetragon are covered.

Smooth interpolation of non-flat Z positions

Claims

Claims 1. A method for determining reference points in automated pipetting system, comprising the steps of: a) defining a coordinate system in the automated pipetting system having two horizontal axes and one vertical axis; b) generating a map of reference points and their nominal position in the coordi-nate system; c) detecting automatically the actual position of the reference points by moving sensors; d) comparing the detected actual position with the nominal position of the reference points; e) determining deviations between the actual and nominal position of the refer-ence points; and 0 adjusting the position of the reference points within the coordinate system corresponding to detected position deviations.
2. The method of claim 1, wherein the sensors are capacity sensors or force feedback sensors.
The method of claim 1 or 2, wherein at least four nominal reference points in the coordinate system are defined and detected.
The method of one of claims 1 to 3, wherein a catch range around each nominal reference point is defined.
5. The method of claim 4, wherein the catch range is maximal 5 mm around a nominal position of a reference point.
The method of one of claims 1 to 5, wherein steps a) to 0 are repeated at least once to ensure consistency and identify errors.
The method of one of claims 2 to 6, wherein the three axes are scanned consecutively.
8. The method of one of claims 1 to 7,wherein the target is grounded when using capaci-ty sensors.
The method of one of claims 1 to 8. wherein sensors of a robotic system of the automated pipetting system are used for detection.
10. The method of one of claims 1 to 9, wherein the actual positions of reference points are interpolated.
11. The method of one of claims 1 to 10, wherein intersecting lines or planes are detected.
12. The method of one of claims 1 to 11, wherein the adjustment of the position of the reference points within the coordinate system depends on predefined thresholds for maximal allowable deviations.
13. The method of one of claims 1 to 12, wherein detected actual position in the vertical axis are fitted in a plane with minimal deviations or alternatively a non-flat surface is generated going through actual positions in the vertical axis.
14. The method of one of claims 1 to 13, wherein the nominal positions are generated as a CAD output or by physically measuring the nominal positions.