WO2023275460A1

WO2023275460A1 - Adaptation of assistance to the presence or absence of effort provided by the cyclist without using a torque sensor

Info

Publication number: WO2023275460A1
Application number: PCT/FR2022/051229
Authority: WO
Inventors: Léonard JAILLET; Julien FLECHE; Raphaël Marguet
Original assignee: Ebikelabs
Priority date: 2021-06-30
Filing date: 2022-06-23
Publication date: 2023-01-05
Also published as: FR3124790B1; FR3124790A1

Abstract

The invention relates to a method for adapting the pedalling assistance on an electrically assisted bicycle, which comprises the steps of: measuring an actual magnitude of movement of the bicycle; measuring the rotation (Vp) of the crankset; calculating a theoretical magnitude of movement of the bicycle from the measurement of the rotation of the crankset; calculating a deviation (∆V) between the actual and theoretical magnitudes of movement; and modifying the assistance when the deviation crosses a threshold (S, S1, S2).

Description

DESCRIPTION

Title of the invention: Adaptation of the assistance to the presence or not of an effort provided by the cyclist without using a torque sensor

Technical area

The invention relates to electrically assisted bicycles, and more particularly to a system for controlling the assistance according to the pedaling activity of the user.

Background

A common category of e-bikes is designed to engage the assist only when the user is pedaling. To ensure optimal comfort of use, the assistance is engaged gradually, often in proportion to the effort provided. This characteristic is ensured in high-end bicycles using a pedaling torque sensor. In more economical bikes, which lack a torque sensor, the motor is often controlled on an all-or-nothing basis based on pedaling activity detected by a rotation sensor mounted on the crankset. The rotation sensor is a simple device comprising, for example, a disc fitted with magnets mounted on the crank axle and a magnetic sensor fixed to the frame. The disc has a multitude of regularly spaced magnets, the number generally varying between 8 and 32, often 12.

To save on the torque sensor while still providing suitable assistance, a number of different methods have been proposed. For example, patent US6152250, in the context of a bicycle equipped with a gear change system, proposes measuring the ratio between the speed of the drive wheel and the speed of the crankset to detect gear changes and modulate the power of support. If the ratio decreases, the cyclist is engaging a low gear to attack a slope - the system then increases the assistance. If the ratio increases, the cyclist is engaging a high gear and needs less assistance - the system then decreases the assistance.

Patent application WO2010/094515 describes a complex system based on the analysis of accelerations over a revolution of the crankset to determine a causal relationship between pedaling and the speed of the bicycle. Summary

There is generally provided a method for adapting the assistance to pedaling on an electrically assisted bicycle, comprising steps consisting in measuring an actual magnitude of movement of the bicycle; measure crank rotation; calculating a theoretical magnitude of movement of the bicycle from the measurement of rotation of the crankset; calculating a difference between the real and theoretical movement magnitudes; and modifying the assistance when the deviation crosses a threshold.

The method may further comprise steps consisting in triggering the assistance when the difference is less than the threshold; and stopping the assistance when the deviation is greater than the threshold.

According to an alternative, the method can further comprise steps consisting in comparing an average of the difference with a first threshold and with a second threshold greater than the first threshold; triggering the assistance when the average of the difference is less than the first threshold; stopping the assistance when the mean of the difference is greater than the second threshold; and decreasing the assistance when the mean of the deviation increases from the first threshold towards the second threshold.

The deviation can be determined by a difference in the magnitudes of movement.

According to an alternative, the deviation can be determined by a ratio of the magnitudes of movement.

The step of calculating the theoretical magnitude of movement may comprise steps consisting in calculating an apparent gear ratio proportional to the ratio of the actual magnitude of movement of the bicycle and of the measurement of rotation of the crankset; and calculating the theoretical magnitude of motion from the apparent gear ratio when the apparent gear ratio satisfies a stability criterion.

The method may further comprise steps consisting in calculating a moving average of the apparent gear ratio over consecutive measurement samples of rotation of the crankset; establish a band around the moving average; deciding that the apparent gear ratio satisfies the stability criterion when a number of consecutive samples of the apparent gear ratio is inside the band; and using an average of the apparent gear ratio to calculate the theoretical magnitude of movement. The number of samples for the calculation of the moving average can correspond to a quarter turn of the crankset, and the number of samples for the stability criterion correspond to a half turn of the crankset.

There is also generally provided an electrically assisted bicycle control system, comprising a first sensor configured to measure the rotation of the motor or of a wheel of the bicycle; a second sensor configured to measure the rotation of a crankset at several regularly distributed points; and a control circuit programmed to implement the aforementioned method based on information from the first and second sensors.

Brief description of the drawings

Embodiments will be set out in the following description, given on a non-limiting basis in relation to the attached figures, including:

FIG. 1 illustrates an example of evolution of the pedaling speed and of a corresponding control quantity used to identify effective pedaling.

FIG. 2 illustrates by a graph a first assistance control mode as a function of the control variable.

FIG. 3 illustrates by a graph a second assistance control mode as a function of the control variable.

Figure 4 illustrates an example of the evolution of an apparent gear ratio calculated from the pedaling speed, in an idealized form.

Figure 5 illustrates the evolution of the apparent gear ratio in a real form corresponding to the example of figure 4.

Figure 6 illustrates the evolution of a moving average calculated on part of the apparent gear curve of Figure 5.

detailed description

In the present application, a system is proposed capable, on an electrically assisted bicycle without a torque sensor, of detecting real pedaling using particularly simple means, requiring no parameterization or sensors other than those already present on a bicycle. entry-level. By "true pedaling", we mean efficient pedaling, providing a non-zero torque, which alone is supposed to trigger the assistance electric - we want by this means to avoid triggering the assistance when the user pedals inefficiently, in a vacuum. Although an absence of pedaling is easy to detect, by an absence of measurement on the pedal sensor over a given interval, it is less easy to detect pedaling in a vacuum whose speed is close to that of efficient pedaling .

The aforementioned patent application W 02010/094515 deals with a similar objective, but describes a complex solution based in particular on an analysis of accelerations during the rotation of the crankset. Moreover, the analysis being made on a certain history of evolution, the solution does not make it possible to react quickly to the transitions between efficient and inefficient pedalling.

In the present application, provision is made to compare an actual magnitude of movement Vr of the bicycle with a corresponding theoretical magnitude of movement Vth based on the current pedaling activity. It is then determined that there is inefficient pedaling when the difference between the actual and theoretical magnitudes becomes too great.

The difference between the quantities can be established in different ways. The embodiments described below use the difference denoted AV, but the ratio could also be used.

The quantities of movement considered can be linear speeds, speeds of rotation, or displacements (linear or in rotation), according to the information most readily available in the control circuit of the assistance of the bicycle. We focus below on the speeds of rotation, by way of example.

The actual speed Vr is generally measured by the control circuit on the basis of information produced by a rotation sensor of one of the wheels or of the motor. The production of the linear speed assumes that the diameter of the wheel is memorized. The theoretical speed Vth can be calculated on the basis of information produced by a pedal rotation sensor, knowing the gear ratio, namely the gear ratio used.

Thus, considering the speeds of rotation:

AV = Vr - Vth, where

Vth = Vp · B (pedaling speed x gear ratio).

On a single-speed bicycle, the gear ratio is fixed and could be memorized in the control circuit. On a multi-speed bicycle, the gear ratio is variable, which implies the implementation of additional means to identify the gear ratio in use. A solution is described later allowing the gear ratio to be measured during use, without prior knowledge of gear ratio values.

Figure 1 illustrates an example of evolution of the pedaling speed Vp and the difference AV noted. The evolution corresponds to a start from a stop situation, beginning with an acceleration phase to reach a substantially constant speed phase.

At a time tO, where the bicycle reaches its cruising speed, the cyclist stops pedaling, but the bicycle continues on its momentum, namely that the real speed Vr practically does not vary. The AV difference, which until then was almost zero, presents a peak. Shortly after, the cyclist resumes pedaling and the AV difference vanishes.

At a time t1, the cyclist slows down his pedaling rate relative to the actual speed of the bicycle, without however stopping pedaling. However, pedaling becomes inefficient. The AV difference has a phase of values greater than zero.

As illustrated, a threshold S can be set which is crossed by the AV difference in each of these illustrated cases to detect inefficient pedaling and adapt the assistance accordingly.

FIG. 2 illustrates by a graph a first possible mode of adaptation of the assistance, in all or nothing. The abscissa axis indicates the value of the AV difference. The y-axis CTRL indicates the motor control rate, the value 1 corresponding to the nominal power.

Thus, as long as the AV difference is less than the threshold S, the assistance is nominal. The assistance is removed as soon as the AV difference exceeds the threshold S.

The low threshold S can be set just above a maximum noise level observed during efficient pedaling. The noise is due to the characteristics of the sensors and the transmission, such as backlash compensation and the elastic effects of the materials involved.

FIG. 3 illustrates by means of a graph a second more progressive mode of adaptation of the assistance. This mode uses a low threshold SI and a high threshold S2, and a moving average of the difference AV, for example over 3 samples, denoted AL. As long as the average DR is lower than the threshold SI, the assistance is at its nominal value. When the average DR is greater than the threshold S2, the assistance is zero. When the average DR evolves from the threshold SI to the threshold S2, the assistance decreases from its nominal value towards 0.

We have represented the progression of the assistance between the thresholds SI and S2 as linear - it could have another decreasing form (exponential, parabolic...).

As previously indicated, on a single-speed bicycle, the gear ratio used to calculate the theoretical speed Vth is fixed and could be stored in the control circuit. On a bicycle with several gears, the gear ratio is variable and it is not enough to memorize the possible gear ratios, because it is also necessary to determine the gear ratio in use.

A technique is described below for measuring the gear ratio in use, without prior knowledge of possible gear ratio values.

This technique is based on an analysis of the ratio of the actual speed to the current pedaling speed Vr/Vp, which is a value corresponding to a gear ratio which will be called the “apparent gear ratio” Ba. Contrary to the usual definition of the gear ratio, according to which the gear ratio is constant for a given gear ratio, the apparent gear ratio Ba varies when pedaling becomes inefficient.

Figure 4 shows an apparent gear change curve based on an example of pedaling activity, assuming that all bicycle drivetrain components and sensors are ideal.

Wherever the crankset is engaged with the drive wheel, namely when the cyclist exerts a non-zero force on the pedals, the apparent gear ratio Ba is equal to the gear ratio in use, or current gear ratio Bc. Changes in the current gear are characterized by stair steps, indicated by A.

In phases indicated by B, the cyclist stops pedalling. The apparent gear ratio becomes infinite.

In phases indicated by C, the pedaling slows down in relation to the speed of the drive wheel, without however stopping. The apparent gear ratio deviates from the current gear ratio upwards, especially as the pedaling slows down. Without prior knowledge of the current gear, it is difficult to know whether phases A, B and C correspond to a variation in pedaling imposed by a change in the current gear or to a voluntary variation by the cyclist in the margin left by the freewheel. Even if the possible gears are known in advance and memorized, it remains difficult to identify a current gear change, because one cannot distinguish a change of the current gear from a start or stop of an inefficient pedaling, all manifested by sudden variations in the apparent gear ratio.

A technique envisaged here for detecting changes in the current gear ratio, or even measuring the current gear ratio, is based, not on variations in the apparent gear ratio, but on the detection of stable phases in the apparent gear ratio. This technique is described below in the context of actual measurements.

FIG. 5 represents a curve illustrating the evolution of the apparent gear ratio Ba as measured in real conditions having substantially the same phases A, B and C as the example of FIG. 4. The scale of the ordinates of the apparent gear ratio corresponds to a unit which turns out to be used in a classic control circuit, namely a development expressed in sixtieths of a meter per revolution of the crankset. This unit produces relatively large integers, which can be processed by a microcontroller without using floating point numbers while ensuring good precision.

The crankset speed sensor here comprises 12 magnets distributed over a disc fixed to the crankset axle, and thus produces twelve pulses per revolution of the crankset. The speed is determined from the time difference between two pulses, which results in twelve speed samples being obtained per crank revolution.

The speed of the driving wheel varies slowly in practice, so the wheel speed sensor is generally designed to produce few samples per revolution, for example only one. When the motor is on the hub, it is the motor speed sensor which can provide the rotational speed of the wheel - in this case the speed measurement generally has better time resolution.

Curve B a includes a calculated point for each sample of the pedal speed sensor. The horizontal axis here representing time, the pitch between the points of the curve Ba varies inversely proportional to the speed of the crankset.

It can be seen that the plateaus of the curve Ba, which correspond to the current gear sought, contain low amplitude noise due to the characteristics of the sensors and of transmission, such as backlash compensation and the elasticity effects of the materials involved.

Pedaling stop phases B are characterized by pulses with different amplitudes. In fact, since the crank speed sensor does not provide a continuous measurement, it is not able to measure a speed tending towards zero. An exception is then applied in the mode of calculation, which produces the slots represented instead of peaks tending towards infinity, which would in fact be unnecessarily disturbing for the subsequent calculations.

Thus, it is considered that a pedaling stoppage occurs when no new sample is produced within a waiting interval, for example 500 ms. The waiting interval can vary inversely proportional to the last recorded pedaling speed. When pedaling restarts, we wait for the second sample to start the new pedaling speed values, which produces limited gear values instead of tending towards infinity.

Phases C of slowed pedaling oscillate at the whim of the cyclist in addition to presenting noise due to the transmission.

There are also values of the curve below the current gear ratio, normally representative of faster pedaling than that imposed by the gear ratio. These values can be explained by the idle travel to be performed to engage the pawls of the freewheel when resuming pedaling after a stop or a change of gear. It can also be back-pedaling when the sensors do not differentiate the direction of rotation.

FIG. 6 is an enlargement of part of the apparent gear curve Ba of FIG. 5, illustrating a technique for detecting stable zones.

A moving average MM is calculated on a number of samples covering a fraction of a crank turn, for example a quarter turn, or 3 samples for a 12-magnet sensor.

Figure 6 illustrates in dotted lines a band defined around the moving average MM. In the example considered, the half-width of the band is equal to MM/40+5, where MM is the current value of the moving average, expressed in sixtieths of a meter per crank revolution. With this configuration, it is decided that the apparent gear ratio satisfies a stability criterion when a given number of consecutive samples of the apparent gear ratio is inside the band, for example 6 consecutive samples corresponding to a half-turn of the crankset. When the stability criterion is satisfied, a moving average of the apparent gear ratio is used as the current gear ratio Bc to determine the theoretical speed Vth used to calculate the AV deviation. The moving average can be the one used for the stability criterion (eg calculated on 3 samples), or a longer moving average, for example calculated on 6 samples.

Moreover, after a certain number of detections of stable phases according to this technique, all of the possible gear ratio values of the bicycle will have been found (each with a tolerance corresponding to the variation band of the stability criterion). These gear ratio values can be stored and updated to follow the average of the measured values. These stored values can be used to accelerate the stability detection phases. For this, for example, each sample of the gear Ba is compared with the set of possible gears on the basis of the stored values. In the event of equality within a tolerance margin, the corresponding memorized gear ratio is selected as the candidate of the current gear ratio Bc. The cases of equality being able to be fortuitous in the event of inefficient pedaling, it will be preferred to confirm the candidate by a succession of two or three consecutive equalities. The current gear ratio for calculating the theoretical speed Vth is thus found in two or three samples instead of the six samples of the stability criterion.

Claims

1. Method for adapting the pedaling assistance on an electrically assisted bicycle, comprising the following steps: measuring a real movement magnitude of the bicycle; measure the rotation (Vp) of the crankset; calculating a theoretical magnitude of movement of the bicycle from the measurement of rotation of the crankset; calculating a difference (AV) between the actual and theoretical magnitudes of movement; and modifying the assistance when the deviation crosses a threshold (S, SI, S2).

2. Method according to claim 1, comprising the following steps: triggering the assistance when the deviation (AV) is less than the threshold; and stopping the assistance when the deviation is greater than the threshold.

3. Method according to claim 1, comprising the following steps: comparing an average of the difference with a first threshold (SI) and with a second threshold (S2) greater than the first threshold; engage the assistance when the average of the difference is less than the first threshold (SI); stopping the assistance when the average of the difference is greater than the second threshold (S2); and decreasing the assistance when the mean of the deviation increases from the first threshold towards the second threshold.

4. Method according to claim 1, in which the deviation is determined by a difference in the magnitudes of movement.

5. Method according to claim 1, in which the deviation is determined by a ratio of the magnitudes of movement.

6. Method according to claim 1, in which the step of calculating the theoretical magnitude of movement comprises the following steps: calculating an apparent gear ratio (B a) proportional to the ratio of the actual magnitude of movement of the bicycle and the measurement of rotation (Vp) of the crankset; and calculating the theoretical magnitude of motion from the apparent gear ratio when the apparent gear ratio satisfies a stability criterion.

7. Method according to claim 6, comprising the following steps: calculating a moving average (MM) of the apparent gear ratio over consecutive measurement samples of rotation of the crankset; establish a band around the moving average; deciding that the apparent gear ratio satisfies the stability criterion when a number of consecutive samples of the apparent gear ratio is inside the band; and using an average of the apparent gear ratio to calculate the theoretical magnitude of movement.

8. Method according to claim 7, in which the number of samples for the calculation of the moving average corresponds to a quarter turn of the crankset, and the number of samples for the stability criterion corresponds to a half turn of the crankset .

9. An electrically assisted bicycle control system, comprising: a first sensor configured to measure the rotation of the motor or a wheel of the bicycle; a second sensor configured to measure the rotation of a crankset at several regularly distributed points; and a control circuit programmed to implement the method of claim 1 from information from the first and second sensors.