Autonomous mobile robot

Abstract

The present invention relates to an AMR having a main body having at least one trailing wheel having a rotation axis, a drive unit for steering and driving the AMR and rotatable relative to the main body around a pivot axis, wherein a steering angle Φ indicates the rotation of the drive unit around the pivot axis relative to a front direction, the drive unit comprising two drive wheels equidistant from the pivot axis, having a common rotation axis perpendicular to the pivot axis, and separated from each other by a distance d along the common rotation axis, wherein each drive wheel has a wheel sensor for measuring the wheel speed vA, vB of the respective wheel, and a steering angle sensor, optionally comprised in the drive unit, for measuring the steering angle Φ, optionally including a rate of change of the steering angle, wherein the rotation axis of at least one support wheel C and the pivot axis are at a distance L relative to each other, and wherein, in a state in which the drive unit drives in the forwards direction, the rotation axis of the at least one support wheel and the common rotation axis are parallel to each other, and a controller for configured to receive from the steering angle sensor a signal representing the steering angle Φ and from the drive unit signals representing the wheel speeds vA, vB of the drive wheels, and evaluate the correctness of the signal representing the steering angle Φ and / or of the signals representing the wheel speeds vA and vB, based on the signal representing the steering angle Φ and the signals representing the wheel speeds vA and vB, by means of the following equation of motion, Formula (I) wherein represents a rate of change of the steering angle Φ.

Description

[0001] ABB Schweiz AG 02.02.2024 A18983WO Autonomous mobile robot Technical field The invention relates to an autonomous mobile robot (AMR). Background Autonomous mobile robots (AMR) may include steerable drive / traction wheels and supporting casters representing a main body of the AMR. As such, the main body follows the speed of the traction wheels, which are typically arranged in a drive unit. The steering angle representing the angle between the drive unit and the main body follows the differential speed of the drive wheels with a time lag. The trailing main body takes time to align with the drive unit after a change of direction of the drive unit. The individual speeds of the traction wheels lead to an average velocity of the AMR. As such, identifying the position and path of an AMR may be challenging. There is a general desire to reduce actuation and sensing on the vehicle. At the same time, information about the vehicle lateral velocity is important to ensure that the correct fields are set for safety laser scanners. Otherwise, the vehicle may run into somebody, simply because its laser scanners are ‘looking’ in the wrong direction. Existing solutions to estimate the actual lateral velocity of an AMR have e.g. the drawback that a check may only be appropriate upon entering a turn for the first time. Known checks may not work on leaving a turn, or during slalom-type manoeuvres, because the main body of the vehicle takes time to align with the drive unit. Thus, it is possible that a check can only be run a few times. Hence, there is a residual risk that an error in the steering angle sensor will not be detected early enough, in particular not before a change of scanner fields is required, which may risk selecting the wrong scanner fields. There is a desire to provide improvements as to the reliability and safety in connection with the movement of an AMR. Summary of the invention The present invention relates to an AMR having a main body having at least one trailing wheel having a rotation axis, a drive unit for steering and driving the AMR and rotatable relative to the main body around a pivot axis, wherein a steering angle Φ indicates the rotation of the drive unit around the pivot axis relative to a front direction, the drive unit comprising two drive wheels equidistant from the pivot axis, having a common rotation axis perpendicular to the pivot axis, and distanced / separated from / by each other at a distance d along the common rotation axis, wherein each drive wheel has a wheel sensor for measuring the wheel speed vA, vB of the respective wheel, and a steering angle sensor, optionally comprised in the drive unit, for measuring the steering angle Φ, optionally including a rate of change of the steering angle, wherein the rotation axis of at least one support wheel C and the pivot axis are at a distance L relative to each other, and wherein, in a state in which the drive unit drives in the front direction, the rotation axis of the at least one support wheel and the common rotation axis are parallel to each other, and a controller configured to receive, e.g. from the steering angle sensor, a signal representing the steering angle Φ and, e.g. from the drive unit, signals representing the wheel speeds vA, vB of the drive wheels (e.g. wherein the signals may be based on measurements of the positions of the wheels, followed by differentiation), and evaluate the correctness of the signal representing the steering angle Φ and / or of the signals representing the wheel speeds vA and vB, based on the signal representing the steering angle Φ and the signals the wheel vA and vB, by means of the following equation of motion, rate of change of the steering angle Φ. The idea underlying the invention is to obtain safe information about the AMR velocity based on the equation of motion and a (direct or indirect) measurement of the wheel speed of two drive wheels as well as a (direct or indirect) measurement of the steering angle. The minimum number of sensors and measured values is three. However, for the velocity of the AMR, two unknown parameters pertaining to the velocity vector arise, as the velocity vector may be defined by a magnitude and a direction, or a longitudinal and lateral velocity at the pivot point of the drive unit. However, decoupling the equations may be less straightforward because the individual wheel speeds vA, vB lead both to an average velocity and also, with some lag, to a steering angle Φ relative to the main body of the AMR. The present invention proposes solutions on this basis for safe velocity determination with only three sensors. The present invention allows for continuous, i.e. permanent, observation of the movement of the AMR. However, while permanent observation is possible, it may not be necessary to indeed run a permanent observation. Hence, reliability and safety in connection with the movement of an AMR is achieved. ISO 3691-4 defines a Performance Level (PL) of c for this functionality for lateral speeds of less than 0.7m / sand a PL of d for higher lateral velocities, which the invention may fulfil.Optionally, as solution 1, the controller is configured to evaluate the correctness by checking the plausibility of the signals by means of a predetermined error threshold of the rate of change of the steering angle , wherein the absolute difference between the rate of change of the signal representing the steering angle Φ, and the rate of change of the steering angle calculated based on the signals representing the wheel vB and the of is above the error This represents a simple and reliable way to increase the safety further. Optionally, as solution 2, the controller is to evaluate the correctness by comparing redundant values with each other, wherein one pair of the pair v’A and v’B and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB, and the other pair of the pair v’A and v’B and the pair v’’A and v’’B corresponds to otherwise obtained wheel speeds of the drive wheels. This offers various measures to increase the safety further by redundancy. Optionally, the wheel speeds v’A and v’B are measured by additional wheel sensors, respectively, and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB. Providing a second, independent measurement for the wheel speeds provides redundancy. Optionally, the pair of wheel speeds v’A and v’B corresponds to the pair of measured wheel speeds vA and vB, and the pair v’’A and v’’B is determined based on a model. Alternatively, the pair of wheel speeds v’A and v’B is determined based on a model, and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB. Providing a model for the wheel speeds provides redundancy. Optionally, the model includes wheel speed demands for the first and second drive wheels, in particular generated based on a programmed behaviour and / or based on adjusting the demanded (commanded) wheel speeds vA and / or vB, based on a low pass filter with an appropriate time constant and / or rate limit. This may represent an efficient and reliable model. In particular, one may use the input to the controller to estimate the speed, but the controller may take a small amount of time to reach the correct setting. Optionally, the comparison includes consideration of a steady state in which the rate of change of the steering angle is zero, and the controller is configured to evaluate the correctness of the signals on this basis before the steering angle Φ is changed. This allows for reliable evaluation before a change of the wheel speeds. Optionally, the controller is configured to determine a prediction for a steady state turn rate of the AMR, based on wherein vLatrepresents a lateral velocity of the AMR measured at the drive unit. Optionally, the controller is configured to evaluate the correctness of the signal representing the steering angle Φ, of the signal representing the wheel speed vAand of the signal representing the wheel speed vB. When a number of measures are implemented at the same time, this may allow for identification of an erroneous signal. Optionally, as solution 3, the controller is configured evaluate the correctness based on an observer comparing the measured steering angle Φ and an estimation of the steering angle Φ including integration of . In control theory, an observer provides an estimate of the internal state of a physical system based on measured inputs and outputs of that system. In the case under consideration, the inputs are the wheel velocities, and the output is the measured steering angle. Optionally, the controller is to obtain the estimation by using numerical integration of without correction from the measured steering angle. This allows for reliable and simple estimation of the steering angle Φ, since the equation is stable for forward motion and steering angles less than 90°. Optionally, the pair of wheel speeds v’Aand v’Bcorresponds to the pair of measured wheel speeds vAand vB, and the pair v’’Aand v’’Bis determined based on a model. Alternatively, the pair of wheel speeds v’Aand v’Bis determined based on a model, and the pair v’’Aand v’’Bcorresponds to the pair of measured wheel speeds vAand vB. Providing a model for the wheel speeds provides redundancy. Optionally, the model includes wheel speed demands for the first and second drive wheels, in particular generated based on a programmed behaviour and / or based on adjusting the demanded wheel speeds vAand / or vB, based on a low pass filter with an appropriate time constant and / or rate limit. This may represent an efficient and reliable model. Optionally, the integration includes a low-pass filter, which is active below a frequency fsplitto provide a low frequency correction from the measured to the estimated steering angle. This makes the method more robust against low frequency drift due to errors in numerical integration and / or wheel slip. Optionally, vAand / or vBare determined based on a model including wheel speed demands for the first and second drive wheels, in particular generated based on a programmed behaviour and / or based on adjusting the wheel speeds vAand / or vB, wherein separate filters with the same time constant are used for the left and right drive wheels. Optionally, a time constant of the filter is different for longitudinal and lateral motions. This may alternatively or additionally improve the estimation of the steering angle Φ, and, thus, the reliability of the error detection. The term “AMR” includes Automated Guided Vehicles (AGVs) and driverless transportation vehicles. Specifically, the present invention relates to a Tug AMR, i.e. an AMR having a trailing axle which simply follows where it is pulled and which may be used for towing trolleys or trailers. Also the invention is directed to a method for controlling an AMR, wherein the AMR is controlled by means of the controller of the invention and / or is controlled as follows: Controlling an AMR, wherein the AMR comprises a main body having at least one trailing wheel having a rotation axis, a drive unit for steering and driving the AMR and rotatable relative to the main body around a pivot axis, wherein a steering angle indicates the rotation of the drive unit around the pivot axis relative to a front direction, the drive unit comprising two drive wheels equidistant from the pivot axis, having a common rotation axis perpendicular to the pivot axis, and separated from each other by a distance d along the common rotation axis, wherein the rotation axis of at least one support wheel and the pivot axis are at a distance L relative to each other, and wherein, in a state in which the drive unit drives in the front direction, the rotation axis of the at least one support wheel and the common rotation axis are parallel to each other, wherein the following steps are carried out: receiving from a steering angle sensor a signal representing the steering angle Φ and from the drive unit signals representing the wheel speeds vA, vB of the drive wheels, and evaluating the correctness of the signal representing the steering angle Φ and / or of the signals representing the wheel speeds vA and vB, based on the signal representing the steering angle Φ and the signals representing the wheel speeds vA and vB, by means of the of motion, rate of change of the steering angle Φ. Further details will be apparent from the detailed description and the drawings, without limiting the invention. Brief description of the drawings Fig.1a shows a schematic top view of an AMR of the invention. Fig.1b shows a side view of an AMR of the invention. Fig.2 shows a diagram of an implementation of the invention. Fig.3 shows a diagram of another implementation of the invention. Fig.4 shows a diagram of another implementation of the invention. Detailed description The invention relates to an AMR 1 having a main body 3 having at least one trailing wheel C having a rotation axis R’, a drive unit 2 for steering and driving the AMR 1 and rotatable relative to the main body 3 around a pivot axis P. A steering angle Φ indicatesthe rotation of the drive unit 2 around the pivot axis P relative to a front direction F. Thedrive unit 2 comprises two drive wheels A, B which are equidistant from the pivot axis P, have a common rotation axis R perpendicular to the pivot axis P, and are separated from each other by a distance d along the common rotation axis R. Each drive wheel A, B has a wheel sensor 4 for measuring the wheel speed vA, vB of the respective wheel A, B, and a steering angle sensor 5, optionally comprised in the drive unit 2, for measuring the steering angle Φ, optionally including a rate of change of the steering angle. The rotation axis R’ of at least one support wheel C and the pivot axis P are at a distance L relative to each other. In a state in which the drive unit 2 drives in the forwards direction F, the rotation axis R’ of the at least one support wheel C and the common rotation axis R are parallel to each other. A controller 11 is configured to receive from the steering angle sensor 5 a signal representing the steering angle Φ and from the drive unit 2 signals representing the wheel speeds vA, vB of the drive wheels A, B, and evaluate the correctness of the signal representing the steering angle Φ and / or of the signals representing the wheel speeds vA and vB, based on the signal representing the steering angle Φ and the signals representing the wheel speeds vA and vB, means of the following equation of motion, rate of change of the steering angle Φ. In other words: The AMR 1 comprises the swivelling drive unit 2 having first A and second B drive (traction) wheels at opposite sides of the swivelling drive unit 2, and one or more trailing / support wheels C in a main body (trailing device) 3. Each drive wheel A, B has its wheel drive unit (not shown) for rotation about the axis R (although the axles are independent, so that the wheels A, B are free to rotate at different speeds). The two drive wheels A, B are preferably independently driven, and the swivelling drive unit 2 is free to rotate about its centre (pivot axis) P by rotation of the two wheels A, B. The support (trailing) wheels C are free of a drive unit and are not driven. The wheel(s) C serve(s) to support the main body 3. If two trailing wheels C are provided, they may have a common rotation axis R’ (but independent axles). The support (trailing) wheels C may be located behind the drive wheels A, B, when seen from the front direction F, and thus, represent a rear axle. In Fig.1a, the rotation axes R and R’ are parallel to each other, when the drive wheels A and B run in the forward direction F. The forward direction F is perpendicular to the rotation axis R’. Put differently, the drive unit 2 drives in the forward direction F, if the drive unit 2 is not swivelled relative to the main body 3. The steering angle Φ of the drive unit 2 relative to the front direction F is measured by a steering angle sensor (encoder) 5, which is optionally located at the drive unit 2. An absolute encoder or corresponding sensor (steering angle sensor) 5 is provided at the drive unit 2 to measure the rotation of the swivelling drive unit 2 relative to the main body 3. There is at least one wheel sensor (drive encoder) 4 on each wheel A, B, or on each wheel drive unit, optionally two sensors 4 per wheel A, B to provide both a speed measurement and motor commutation. Each wheel A, B may be controlled by a motor drive (not shown) which implements velocity commands from a vehicle guidance system. The kinematic equation for movement of the AMR represents the basis of the method, the wheel speeds of wheels A and B, respectively, and is the steering angle. The wheel speed v refers to the translation velocity of a wheel. A distance L (wheelbase) is between the pivot axis P of the drive unit 2 and the trailing axle (rotation axis R’ of the trailing wheel(s) C). A distance d is between the two drive wheels A and B, as shown in Fig.1a. Fig.2 shows a side view of the AMR 1, wherein the steering encoder (steering angle sensor) 5, the drive encoder / wheel sensor 4, trailing device (main body) 3 and the (swivelling) drive unit 2 are shown. The rotation axis R of drive wheels A, B and the rotation axis R’ of support wheel C run into the plane depicted in Fig.1b. The forward or front direction F is indicated in Fig.1a and 1b, and is perpendicular to the rotation axes R, R’ and the pivot axis P. By using this equation of motion, it is possible to check and evaluate whether an error has occurred in connection with the velocity measurement of the AMR, e.g. to cross- monitor the sensors 4, 5 at all times during operation, which may improve the likelihood of detecting a fault. According to solution 1, the is re- such that a fault is detected if Here, the controller 6 is configured to check the plausibility of the signals by means of a predetermined threshold of the rate of change of the steering angle Φ, wherein the absolute difference between the rate of change of the signal representing the steering angle Φ, and the rate of change of the steering angle calculated based on the signalsrepresenting the wheel speeds vA, vB and the equation of motion is above a predetermined threshold . This may be considered to be equivalent to a test function for a Category 2 architecture in ISO 13849-1. There is no redundancy. All three sensors are needed to produce a valid comparison. Solution 2 may allow for generating redundant measurements and in particular to produce a Category 3 architecture according to ISO 13849-1. SRP / CS of category 3 may be designed so that a single fault in any of these parts does not lead to the loss of the safety function. The MTTFd of each of the redundant “channels” may be low-to- high. It may be easier to achieve a higher safety performance level with this architecture, which is why redundant measurements are preferred. The kinematic equation is re-arranged to move all the angle measurements to one side, namely the left side. The two sides of the equation represent the two channels, wherein the wheel speeds vA and vB are, for the sake of generalization, replaced by wheel speeds v’A and v’B and v’’A and v’’B are follows: wherein one pair of the pair v’A and v’B and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB, and the other pair of the pair v’A and v’B and the pair v’’A and v’’B corresponds to otherwise obtained wheel speeds of the drive wheels (A, B). There are three possibilities to make the channels (1) and (2) redundant (i.e. independent): 1. Add a second velocity sensor 4’ on each drive wheel A and B, so that one measurement can be used for the left-hand side of the equation, and the other measurement can be used for the right-hand side of the equation. A drawback may be that two additional sensors 4’ are required. However, it might be cheaper than adding e.g. a second steering angle sensor with sufficient resolution and accuracy. This means, with respect to channels (1) and (2) above, that for channel (1), the wheel speeds v’A and v’B are measured by additional wheel sensors 4’, respectively, and for channel (2), the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB measured by the sensors 4. This possibility may be convenient for vehicles where a second sensor 4’ is already provided at each drive wheel A, B. 2. Use the measured values on the left-hand side of the equation (i.e. channel (1)) and generate the values on the right-hand side (i.e. channel (2)) from a model. Here, the prediction channel (1) is separated fromthe measurement channel (2). This means, with respect to channels (1) and (2) above, that for channel (1), the pair of wheel speeds v’A and v’B corresponds to the pair of measured wheel speeds vA and vB, and for channel (2), the pair v’’A and v’’B is determined based on a model. 3. Generate the wheel speeds on the left-hand side (channel (1)) of the equation by using a model, while using the measured wheel speeds on the right-hand side (channel (2)). This mixes prediction and measurement “channels”, but this may not be a disadvantage from the safety perspective.This means, with respect to channels (1) and (2), that the pair of wheel speeds v’Aand v’B is determined based on a model, and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB. Also detection of the velocity before the next demand is carried out may be desired. In the steady state, which may be assumed to be present before a change of the demand, , which means: Here, each of the channels represents a prediction of the steady state vehicle turn rate. This is directly proportional to the vehicle lateral velocity (multiply by L). This approach may be seen as looking ahead of the actual motion and may allow to detect a fault before the next demand on the safety function (= correct switching of lateral fields). This may allow to diagnose faults in any one of the sensors (wheel speed vA and vB and steering angle Φ, e.g. split into an angle and rate measurement). In principle, any single fault in the measurement system can be detected. As such, the controller is configured to evaluate the correctness of the signal representing the steering angle Φ, of the signal representing the wheel speed vA and of the signal representing the wheel speed vB. Solution 3 relies on a mathematical model to generate an estimate of the steering angle and is described as follows: An observer is based on the kinematic equations of motion. This is shown diagrammatically in Fig.2. This directly implements the kinematic equation to produce an estimated steering angle based on the measured velocities vA and vB of the two wheels A and B, respectively, only. The -block represents an integrator. The right- hand side of the kinematic equation is integrated and then compared with the measured angle , yielding which indicates a deviation / error in the steering angle. A model is used for the speeds of the wheels A, B. This may be considered a variation of an aspect of the solution 2. In both solutions 2 and 3, one may take the demand to the motor control, and pass it through a model (optionally a first order filter, further optionally with a rate limit) and then use the output to represent the actual wheel speeds. In solution 2, one may use the same filter on each wheel, and then use the resulting speeds to calculate the longitudinal and lateral motion. In this case, one may still take two filters, but put them on the average wheel command and the "difference" wheel command. In particular the steering angle may change faster than the vehicle speed because of the different inertia, so this may more precisely model the vehicle behaviour. For solution 3, one may either use separate filters with the same time constant for each drive wheel, or split the terms into an average and a difference as described. Hence, the controller is configured to evaluate the correctness based on a comparison of the measured steering angle Φ and an estimation including integration of An advantage of this approach may be that integration is a smoothing process, so the equation acts as a filter on sensor noise, which actually corresponds to the mechanical filter in the motion of the AMR itself. The equation itself is stable, so even if the initial steering angle Φ is incorrectly initialised, the estimate should converge to the actual value Φ. This may be evident from the following: Assuming a small angle, this yields: Using the Laplace notation, this The eigenvalue of this equation Hence, this is clearly stable (negative) if the average speed is positive, which is the normal operating case for the AMR. This may only break down when the gradient of the sine function becomes negative at >90°. In that case, the equation only remains stable with a negative drive unit speed – i.e. the vehicle is still moving forward. Should there be problems with numerical drift or initialisation, the observer may be split into low and high frequency components. Here, a low frequency correction term is introduced, which is active below a frequency of fsplit(Hz). This is represented by the summing junction on the left of Fig.3, and the gain . Hence, the integration includes a low-pass filter to implement this correction, which has a bandwidth equal to the frequency fsplit. This means that the fault detection may only function correctly above the frequency of the low pass filter. However, above this decoupling frequency, there may be a direct comparison between the measured and the predicted steering angle. If the error exceeds a defined threshold, a fault is detected in the steering angle sensor. Like in solution 2, it is also possible to use modelled wheel velocities vA, vBto implement this observer, as described above. An example implementation is shown in Fig.4, based on modelled wheel speed of first / second wheel vADem, vBDem. This may decouple the steering angle monitoring from the wheel speed measurements. The low pass filter includes a first order filter . A first order filter may be used to represent the response of the wheel speed control loop. Higher order filters or models of the controller and actuator or individual filters for each wheel are conceivable. The time constant τ of the filter may be different for longitudinal and lateral motion. E.g. τ may consider the acceleration of the whole AMR for longitudinal (symmetric) movement, while for lateral (antisymmetric) motion the acceleration of the pivoting drive unit may be considered, see τSym, τ Asym. The latter may be based on the inertia present in the drive unit 2, which is constant and (relatively) low. The symmetric time constant may depend on the mass of the vehicle, which is higher, together with towed mass, which may be variable. The solutions 1, 2 and 3 of the invention may be used in combination with one or two other ones of the solutions 1, 2, 3.

[0002] Reference signs 1 AMR 2 drive unit 3 main body 4 wheel sensor 5 steering angle sensor 6 controller A first drive wheel B second drive wheel C support wheel(s) F forward or front direction R rotation axis of drive wheels R’ rotation axis of support wheel(s) P pivot axis vA measured wheel speed of first wheel vB measured wheel speed of second wheel d distance between drive wheels A and B L distance between pivot point P of drive unit and rotation axis R’ of trailing wheel(s) v’A / v’B / v’’A / v’’B wheel speed of first / second wheel vADem, vBDem modelled wheel speed of first / second wheel vLat lateral AMR velocity fsplit threshold frequency for low pass filter τ, τSym, τ Asym time constant, symmetric and antisymmetric Φ steering angle measured steering angle estimated steering angle deviation / error in steering angle rate of change of the steering angle threshold of rate of change of the steering angle

Claims

Claims:

1. AMR (1) comprising: a main body (3) having at least one trailing wheel (C) having a rotation axis (R’), a drive unit (2) for steering and driving the AMR (1) and rotatable relative to the main body (3) around a pivot axis (P), wherein a steering angle Φ indicates the rotation of the drive unit (2) around the pivot axis (P) relative to a front direction (F), the drive unit (2) comprising two drive wheels (A, B) equidistant from the pivot axis (P), having a common rotation axis (R) perpendicular to the pivot axis (P), and separated from each other by a distance d along the common rotation axis (R), wherein each drive wheel (A, B) has a wheel sensor (4) for measuring the wheel speed vA, vB of the respective wheel (A, B), and a steering angle sensor (5) for measuring the steering angle Φ, wherein the rotation axis (R’) of at least one support wheel (C) and the pivot axis (P) are at a distance L relative to each other, and wherein, in a state in which the drive unit (2) drives in the front direction (F), the rotation axis (R’) of the at least one support wheel (C) and the common rotation axis (R) are parallel to each other, and a controller (11) configured to receive a signal representing the steering angle Φ and signals representing the wheel speeds vA, vB of the drive wheels (A, B), and evaluate the correctness of the signal representing the steering angle Φ and / or of the signals representing the wheel speeds vA and vB, based on the signal representing the steering angle Φ and the signals representing the wheel speeds vA and vB, by means of the ofrate of change of the steering angle Φ.

2. AMR of claim 1, wherein the controller (6) is configured to evaluate the correctness by checking the plausibility of the signals by means of a predetermined error threshold in the rate of change of the steering angle , wherein the absolute difference between the rate of change of the signal representing the steering angle Φ, and the rate of change of the steering angle calculated based on the signals representing the wheelspeeds vA, vB and the equation of motion, is above the predetermined threshold , namely3. AMR of claim 1, wherein the controller is configured to evaluate the correctness by comparing redundant valueseach other, wherein one pair of the pair v’A and v’B and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB, and the other pair of the pair v’A and v’B and the pair v’’A and v’’B corresponds to otherwise obtained wheel speeds of the drive wheels (A, B).

4. AMR of claim 3, wherein the wheel speeds v’A and v’B are measured by additional wheel sensors (4’), respectively, and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB.

5. AMR of claim 3, wherein the pair of wheel speeds v’A and v’B corresponds to the pair of measured wheel speeds vA and vB, and the pair v’’A and v’’B is determined based on a model.

6. AMR of claim 3, wherein the pair of wheel speeds v’A and v’B is determined based on a model, and the pair v’’A and v’’B corresponds to the pair of measured wheel speeds vA and vB.

7. AMR of claim 5 or 6, wherein the model includes wheel speed demands for the first and second drive wheels, in particular generated based on a programmed behaviour and / or based on adjusting the wheel speeds vA and / or vB.

8. AMR of any of claims 3 to 7, wherein the comparison includes consideration of a steady state in which the rate of change of the steering angleis zero, and the controller is configured to evaluate the correctness of the signals on this basis before the steering angle Φ is changed.

9. AMR of claim 8, wherein the controller is configured to determine a prediction for a steady state turn rate of the AMR, based onwherein vLat represents a lateral velocity of the AMR.

10. AMR of claims 3 to 9, wherein the controller is configured to evaluate the correctness of the signal representing the steering angle Φ, of the signal representing the wheel speed vA and of the signal representing the wheel speed vB.

11. AMR of claim 1, wherein the controller is configured evaluate the correctness based on an observer the measured steering angle Φ and an estimation including integration12. AMR of claim 11, wherein the integration includes a low-pass filter, which is active below a frequency fsplit, in particular to incorporate a low frequency correction of the estimated steering angle to the measured steering angle Φ.

13. AMR of claim 11 or 12, wherein vA and / or vB are determined based on a model including wheel speed demands for the first and second drive wheels, in particular generated based on a programmed behaviour and / or based on adjusting the wheel speeds vA and / or vB, wherein a time constant of a filter in the model is optionally different for longitudinal and lateral motions.

Images