Personal vehicle, and control apparatus and control method therefore

Description

TECHNICAL FIELD

The invention relates to a personal vehicle, in particular to an electric wheelchair, which can be controlled by an inertial body-mounted motion sensor, and a control device and method for controlling the wheelchair.

STATE OF THE ART

[0002] The control device disclosed in WO 2007/082969 A1 includes a head mounted wireless transmitter, a microcomputer, as well as accelerometric and gyroscopic micro-sensors adapted to measure the angular position of the head in both vertical and horizontal directions. A computer or electric wheelchair can be controlled by the measured angular data. However, this known solution has the problem that it employs only a single sensor module which measures the combined tilt angle of the wheelchair and the user's head and allows the vehicle to be steered solely by tilting the head. The following disadvantages are associated with this:

- In a real city environment, a head-mounted motion sensor detects the combined movements of the vehicle and the user, and thus the disturbance from vibrations caused by uneven pavement is added to the vehicle's control signal. Due to these disturbances, the control system performs undesirable actions to steer the vehicle, including sudden turns and unexpected changes in speed.

- When only a single head-mounted sensor is used, the maximum vehicle acceleration may not be used as in the event of sudden starts and stops, the sensor detects high accelerations due to head inertia, resulting in oscillations between sudden starts and stops during wheelchair movement.

- On slopes and wheelchair ramps, an angle sensor mounted on the head measures the deviation of the entire steering-vehicle system. When going uphill, the user has to tilt his head more and more forward to prevent the vehicle from losing speed and eventually coming to a stop. However, when driving downhill, if the user tilts his head back more and more, the vehicle may accelerate excessively.

- In the event that the head-mounted sensor moves for some reason, the relationship between the measured head orientation and the vehicle's control signals will not be accurate. As the user also needs to learn how to steer the wheelchair using the controller (which is not easy to position correctly), the slightest movement of the sensor may prevent proper vehicle control. In extreme cases, the sensor can even fall off the head, which can lead to completely uncontrolled movements.

- In many cases, paralysis affects some of the muscles in the neck rather than any part of the limbs. In such cases, the head-mounted sensors cannot be used.

- This head movement detection solution uses only accelerometric sensors to measure the head tilt angle. In an indisputable proportion of cases, the user is only able to tilt the head to a certain extent, but cannot easily turn it.

[0003] The device disclosed in US 2008/0048931 A1 is adapted to target projectiles fired from an airplane by measuring the angular position of the pilot's head with respect to the airplane. This solution combines eye and head tracking with an infrared camera with signals from accelerometers and gyroscopes. This known solution has the disadvantage that infrared computer recognition is an expensive and resource-intensive technology that is only reliable in an environment in which it is well controlled - such as in an airplane cockpit environment. This document does not disclose the use of integrated multi-sensor signals.

[0004] The device disclosed in US 2012/0143400 A1 comprises a wireless head mounted transmitter, a microcomputer as well as accelerometric and gyroscopic micro-sensors adapted to measure the angular position of the head in both the vertical and horizontal directions. According to

ΕΡ 3 076 906 Β1 as described in the documentation, the head tilt angle is proportional to the wheelchair speed, and the head tilt direction is the same as its direction of movement. This method has the same disadvantages as the control method of WO 2007/082969 A1, and does not provide for other relations describing the input and output signals.

[0005] In the system according to US 2012/0310470 A1, control information and diagnostic data are received and sent by a control unit of the wheelchair via a short-range wireless connection, where the short- and long-range wireless transmitter can also be a smartphone or a tablet. The disadvantages of this solution are that the smartphone or tablet can only be connected to the wheelchair using a wireless connection, which is less secure than wired connections, and that the built-in sensors of the smartphone or tablet are not used to control the vehicle.

[0006] According to US 2008/0135321 A1, at least one sensor adapted to detect an angular yaw of the vehicle is mounted on the electric personal vehicle, and overturning or blocking of the vehicle is automatically signaled by this sensor. This has the disadvantage that the sensor signals are not used to control the vehicle.

[0007] The solution disclosed in US 2013/0253769 A1 prevents the cart from drifting away from its direction by monitoring the angular velocity of the wheels using optical or magnetic sensors and compensating it depending on the signal received. This has the disadvantage that most wheelchairs do not have an encoder. Sensors and relays located on or near the wheels are susceptible to soiling and mechanical damage.

[0008] US 5,555,495 describes the so-called the "man in the loop" method, in which the person is treated as a disturbing variable in the human-machine system. Matching is implemented through model building, which is disadvantageous because it has a slow learning curve and does not develop the user's skills and abilities (which is of great importance for a person in a wheelchair). This requires resources and can only learn from the instructions already carried out, which in some cases is inadequate.

DISCLOSURE OF THE INVENTION

[0009] The object of the invention is to eliminate the drawbacks of the known solutions and thereby to provide a personal vehicle and a vehicle control device and method that allows safe and user-adaptable driving control. Another object is to provide control that can filter out and automatically compensate for disturbances due to uneven road surfaces, uphill and downhill surfaces, and thus improve driving stability.

[0010] The above objects are achieved with a control device according to claim 1, a personal vehicle according to claim 7 and a method according to claim 9. Preferred embodiments of the invention are defined in the dependent claims.

BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Preferred embodiments of the invention are described below with reference to the following drawings in which

Fig. 1 is a schematic three-dimensional drawing of an exemplary electric wheelchair provided with a control device according to the invention,

Fig. 2 is a block diagram of exemplary motion detection units used in the invention,

Fig. 3 is a schematic drawing of a head-set that may be used with the invention, Fig. 4 schematically illustrates the gestures performed by the head for steering,

Fig. 5 schematically illustrates the interconnection of an exemplary control device implemented as a hardware / software platform,

Fig. 6 is an illustration of three levels of user customization,

Fig. 7 schematically illustrates possible platform entrances,

Fig. 8 schematically illustrates the use of two different types of input devices,

ΕΡ 3 076 906 Β1

Fig. 9 schematically illustrates the measures taken to control stability,

Figures 10A and 10B show a schematic view of the structure of a preferred control device according to the invention and the operation of the device, i

Fig. 11 shows a graphic displayed on an exemplary user interface to position the headset on the head and the motion detection unit associated therewith.

WAYS TO CARRY OUT THE INVENTION

[0012] In Fig. 1 a three-dimensional drawing of an exemplary electric wheelchair 10 provided with a control device according to the invention is shown. The invention can of course be used in conjunction with other personal vehicles, but an example with a wheelchair is described in more detail below.

[0013] In the illustrated preferred embodiment, the wheelchair 10 can be controlled, in addition to using a conventional joystick 14, by means of a so-called inertial sensors and body-mounted motion sensors. According to the invention, at least two motion detection units and a processing unit are used in the following arrangement: the first motion detection unit 11 is attached to the freely moving part of the rider, and the second motion detection unit 12 is attached to the wheelchair 10, i.e. to the personal vehicle. . Both motion detection units use the environment as a reference base on which the vehicle and driver are positioned. In a preferred embodiment, the datum, i.e. the common frame of reference, is defined by the direction of gravity and a direction of magnetic (e.g. magnetic north). As to the orientation, the two directions define a unique reference frame in space, the orientations may be determined by the two motion detection units 11, 12. The orientations of the two motion detection units 11, 12 change continuously due to the movement of the units. In the context of the present application, the expression "motion detection unit" is used to denote any units capable of determining a yaw and a roll value (hereinafter: orientation) with respect to an external reference system. As described below, the relationship between the orientation of the vehicle and the rider is computed by a processing unit using the combined signals of the two motion detection units 11, 12, and these relationships are then used by the processing unit to control the vehicle.

[0014] In a preferred embodiment of the wheelchair control platform which is also controlled by inertial sensors, the motion sensors may be any electronic devices that are capable of detecting the orientation in the 3 axis system. The orientation sensor mounted on the user may be attached to any part of the body. In the illustrated preferred embodiment, the first head mounted motion detection unit 11 is shown. As shown in Fig. 1, The platform electronics have a wireless connection with the tablet and inertial headphone kit.

[0015] According to the invention, the first and second motion detection units 11, 12 are adapted to define an orientation with respect to the same external reference frame. Based on the first orientation provided by the first motion detection unit 11 and the second orientation provided by the second motion detection unit 12, a so-called An inertial control signal sent from the processing unit 13 as a function of the relative orientation of the first motion detection unit 11 and the second motion detection unit 12. In a preferred embodiment of the invention, the processing unit 13 is implemented as a PC, tablet, smartphone or game console.

[0016] The control device according to the invention preferably observes the trends of changes in the input parameters and compares the schemes. It is also capable of self-learning and recognizing gestures that can be assigned randomly selected functions. In an "intelligent" embodiment, the control device is capable of learning gestures and driving habits, and thus is able to adapt both to the surroundings and to the driver of the vehicle.

[0017] In a more preferred embodiment, the control device is also capable of remotely monitoring the personal vehicle. It is further advantageous if the user can use the control device to control a computer, smartphone or tablet or to send a signal to electronic switches for lighting, doors, elevators and other electronic devices.

EP 3 076 906 Β1

[0018] The control signal generated due to the difference between the orientation of the body parts moved by the user and the orientation of the wheelchair 10, or in other words, their relative orientation has been termed "inertial control signal". The reason for arranging, in the first preferred embodiment, the first motion detection unit 11 in the head-mounted assembly (head set) is that the invention was developed primarily for paralyzed persons below the neck.

The external reference system (reference base) is therefore preferably a coordinate system defined by the magnetic field and the gravity of the earth. When using such a reference system, commercially available sensors, e.g. magnetometers (comparison to the magnetic field of the earth) and accelerometers (comparison to the gravitational field) can advantageously be used. In Fig. 2 the structure of the first and second motion detection units 11, 12 which are identical is illustrated. Each of the units comprises a fast response three-axis gyro sensor 20, a long response three-axis magnetometric rotation sensor 21 adapted to recalibrate the signal from the gyro speed sensor 20 periodically, and a three-axis accelerometer 22 adapted to compensate for the drift of the gyro speed sensor 20. These sensors preferably consist of components in the form of one-dimensional sensors oriented in the direction of three axes (frame of reference). It is clear that the same sensors can also be used to detect the directions defining the reference system, which facilitates setup.

[0020] The term "sensor merging" is used in the present invention in two ways. First, during inertial control, control is performed based on the relative (reciprocal) orientation of the two motion detection units - preferably of identical structure. Secondly, the driving stability of the wheelchair 10 is provided by motion sensors located in the control device itself. As will be described below, drift is compensated according to the invention on the basis of gravity and geomagnetism.

[0021] Due to the vibration effect, the micro-electromechanical gyros (best suited for measuring rapid movements) within the three-axis gyro rotation sensor 20 are particularly sensitive to the phenomenon of "drift". The so-called gyro drift corresponds to the case where the output sensor reading is non-zero at rest. This undesirable effect - false rotation detected without simultaneous acceleration - is recognized by the insanely reliable accelerometers contained in the triaxial accelerometer 22 and compensated for by the feedback of the PID controller 23 controlled by the difference value, shown in Fig. 2. For sensor error compensation and gyro error recognition, the signals are normalized.

A solution using gyroscopes and compensated by accelerometers becomes more stable by orders of magnitude when using magnetometer signals (which respond slowly but free from drift errors) to periodically (i.e. dynamically) recalibrate the gyro sensors according to with angular position provided by magnetometers. The time difference between recalibrations may be selected e.g. empirically depending on the particular application. Without recalibration, sensor measurement errors would accumulate when the sensors were merged, which would result in a slow rotation (drift) of the direction vector. The compensation process thus involves merging the data from the long-response magnetometer and the fast-response gyro. The raw gyro and magnetometer data are corrected in proportion to the motion independent drift value appearing on the integrated output reading from the three sensors. In extreme cases, the gyroscope may overload (lock up) when - in the absence of any acceleration other than that of gravity - the gyro electronics must be reset after the magnetometer is set in the right direction. Thus, the static rotation (i.e., rotation in the absence of non-gravity acceleration) of the direction vector obtained from the complex sensors is measured during operation and the raw data is compensated proportionally to the error value.

[0023] The described method is performed on the signals from three different sensors for all axes corresponding to the dimensions in space, vectors are added and the orientation obtained in this way is processed preferably in the quaternion projection.

ΕΡ 3 076 906 Β1

[0024] In addition to the triaxial accelerometer 22, the triaxial gyro speed sensor 20 and the triaxial magnetometric speed sensor 21, as well as the microprocessor and the radio transceiver, two additional sensors may be placed in the first motion detection unit 11. As shown in Fig. 3, a pressure sensor 30 may be provided at the back of the headband to allow the user to send a mode change command to the controller, and an infrared distance sensor 31 adapted to detect the blinking of a person wearing the headband.

[0025] When a sudden change is detected in the pressure between the headband and the wheelchair headrest 10 - i.e. the user's head has been tilted back - the operating mode of the wheelchair controller is changed by the pressure sensor 30. Two basic sensor functionalities This pressure is controlling the wheelchair itself and controlling the wheelchair user interface by moving the head. Communication and environmental control functions can also be obtained via the user interface. Pressing the pressure sensor 30 while driving may e.g. automatically stop a wheelchair or start a user interface.

[0026] The distance sensor 31 located near the user's eye uses the physical characteristics of the human eye, i.e. that the vitreous body absorbs light, the eyelids reflect it, and thus different values ​​can be measured in the open and closed positions of the eye. The blink frequency in the vegetative state shows a correlation with the user's wakefulness, and more importantly, blinking can be used as an "input device" to complement a conventional mouse by assigning different functions to different blink durations. For example, a blink between 500 ms and 1 s may correspond to pressing the left mouse button.

[0027] As sensor fusion is performed for the head-set motion sensor in the same way as for wheelchair-attached sensors, a signal can advantageously be generated from the difference therebetween, which can be used for control purposes. Four smooth point values ​​are required to represent each vector numerically: the angles in each of the three planes in the space between the gravity vector used as the reference and the mapped vector (pairs of angles corresponding to the current plane of rotation) and the angle of rotation of the vector (about its axis) relative to magnetic north as reference. When the control device is first actuated, the orientation of the wheelchair measured with respect to the reference directions (gravity, magnetic north) is linked to the driving direction of the wheelchair which will be used as the reference for the control signal.

[0028] As the user puts the headset on and off each day, when the wheelchair is actuated, the orientation of the headset on the head with respect to the external reference system is linked to the forward orientation. Then, in a manner similar to joystick devices, only angular differences with respect to the reference wheelchair 10 are processed.

[0029] According to the default settings, the proportional setting of the speed of the vehicle can advantageously be made by tilting the head backwards / forwards as shown in the left part of Fig. 4. In the default situation illustrated in the right part of this figure, the turn is preferably achieved by tilting or turning your head left or right. The default settings can of course be changed at the user's request. The primary functionality of the control device - implemented as a hardware / software platform in the embodiment shown in Fig. 5 - is to provide safe control and guidance of the wheelchair 10, including filtering and normalizing control signals, assigning signals to specific actuation functions, as well as providing driving stability. It is especially important to filter out external disturbances caused by uneven road surfaces and to compensate for the drift that occurs when turning and driving straight.

[0031] In addition to joystick control and Inertial Measurement Units (IMUs), the control device may allow other special, secondary control modes, for example taking into account alternatively controllable inputs. Additional sensors used for control can be, for example, distance sensors, strain sensors, strain gauges, touch panels, mice, position sensors or stereoscopic cameras. Thanks to specialized

ΕΡ 3 076 906 Β1 due to the functionalities that can be obtained thanks to environmental sensors, automatic obstacle avoidance also becomes possible.

[0032] Additionally, the control device may provide a wireless connection with a computer, mobile devices operated by the user with gestures used to control the wheelchair. The system configuration can also be done over a wireless connection using a tablet or smartphone, which can also be used as a display and communication device.

[0033] Moreover, the control device may issue control commands to door openers, light switches, electronic devices, and thus perform environmental control and command functions. There is no product on the market that would provide all of the above functionalities.

[0034] The platform is preferably divided into three well-separated layers, the "lowest" execution layer being the electronics power unit and its microprocessor, in other words the power module, which is directly responsible for controlling the motors and brakes.

[0035] The next layer above it is the so-called A "lower level layer" to which various input devices are connected communicating with different protocols. This layer includes a 3x3 axis motion sensor, radio and infrared transmitters and receivers, and a microprocessor that performs signal filtering and control signal calculation. To ensure safe operation, all basic control functions of the wheelchair are implemented in this layer.

The uppermost layer, which is the user interface, preferably comprises a smart mobile device with built-in GSM, GPRS and GPS modules, a touch screen, a camera, sound processing devices and speakers. This device can be used to configure the entire platform, but is not necessary for its operation. The settings available on the interactive user interface range from engine power limitation to wheel diameter, personal physical parameters (weight, height), and ranges of motion for filter selection and tuning. In addition to this, since the bottom layer allows you to monitor the stable operation of the mobile device, the platform also allows you to run - on a strong device - learning algorithms, modeling systems, motion recognition applications or software capable of intelligent navigation. As soon as the bottom layer detects that the response time of the smart device has increased (e.g. noACK signal received by the time expired), the top layer is excluded from vehicle control. However, all such control-related parameters that increase driving safety are stored in the lower level layer, ensuring adequate control even in the absence of a smart device.

The versatility of the control device is ensured, above all, by its adaptability. When controlling a wheelchair, it is most important that the control system provides the highest possible level of safety in line with the driver's abilities.

[0038] Fig. 6 illustrates three levels of adjustment. The lowest level, adapted steering, means that within the safety limits of the motors (overcurrent protection), the degree of acceleration, the degree of the so-called "Ramp" can be implemented by the user according to his needs. There is currently no device on the market that allows wheelchair owners to make such changes. The system advantageously allows for a plurality of pre-programmed "behaviors" from which the user can select whichever is most appropriate to the driving conditions, for example low power consumption - long range, or quick maneuvering.

[0039] The component shown at the center of the figure, i.e. the customizable signal filtering component, is used to remove muscle jerks, occasional seizures as well as uncertainty due to fatigue due to muscle weakness and convulsions with great efficiency. Software filters - based on clinical tests - are preferably selected according to different types of problems. The parameters of the simplest low-pass, high-pass, window, linear or recursive methods are set by the controller in an adaptive manner and are thus capable of filtering out long-term disturbances (such as muscle fatigue).

ΕΡ 3 076 906 Β1

[0040] Customizable inputs mean compatibility with several input devices and the ability to configure such devices. In a basic embodiment of the controller, the device is a 3x3 = 9 axis motion sensor attached to the body part. However, for people with disabilities who are able to move their limbs, driving a wheelchair should be an activity combined with the simultaneous exercise of healthy parts of the body. In this way, the control device may be factory-integrated with other types of input devices than those described above, for example optical or capacitive joysticks or touch panels. The control device may include several standard connection interfaces (for example, USB, BT, WFI, IRDA).

[0041] As shown in Fig. 7, the platform may include several inputs on both the microprocessor unit and the mobile smart device. Since each person has a different capability profile, each driver of the wheelchair 10 should be provided with the most suitable input device. In extreme cases, a user completely paralyzed below the neck (all four limbs affected) can steer the wheelchair with simple head movements and spatial head movements. The physiological causes of diplegia can include many medical cases. People suffering from chronic muscular dystrophy who gradually lose muscle strength may only use special input devices with very high sensitivity, such as capacitive touchpads or optical joysticks. The intention to configure the platform as described above was to ensure compatibility with the largest possible number of devices available on the market. Patients with neurological disorders affecting other organs except the locomotor system (e.g. senses, sight) can be equipped with devices supporting navigation using optionally connected environmental sensors.

[0042] Following the sequence of steps illustrated in Fig. 8, basically two basic types of input devices can be used in a controller: joystick devices (measuring absolute displacements) and mouse devices (measuring relative displacements).

In the case of the former, after the peak values ​​have been measured (calibrated), the signal is adjusted to a range compatible with the system (normalization), and then entered into the so-called a dispatch module 35, which is a software module adapted to assign input signals to output signals corresponding to a configuration.

[0044] The mouse devices provide information on the relative displacement, so that peaks are assigned ranges of variation and then used to model the physical operation of the joystick (physical characteristics of the model such as spring force can be calibrated). For example, using a touchpad or sliding a finger across the sensor provides a vector that starts from anywhere. In this case, however, calibration of the relative displacements is required. As long as the finger touches the sensor surface, the Cartesian coordinates characteristic of the physical joystick are transmitted by a program module adapted for joystick modeling and emulation, and then entered into the shipping module 35 after normalization. However, as soon as the finger is lifted from the sensor surface, the virtual joystick returns to the center position thanks to the joystick emulator using physical spring models.

[0045] In addition to the above-mentioned two types of signals, further signals are also present, for example signals from switches and environmental sensors - which are not necessarily directly related to the control of the wheelchair (e.g. horn). These signals are transmitted to the shipping module 35 after normalization.

[0046] Fig. 9 schematically illustrates the measures taken for stability control. Personal vehicle "drift" is a phenomenon that consists in deviating from the direction of travel set by the user. It may be the result of uneven road surfaces, the difference between the holders of the drive wheels on the left and right sides or the distance traveled by them (for example on a slope), or the uneven operation of the drive motors and differences due to mechanical wear.

[0047] To compensate for vehicle drift, a PID controller known per se may be used, the process variable being the integrated data of the inertial sensors of the platform - vehicle orientation, and the control signal of the vehicle being the controlled variable. The software component of the PID controller calculates the difference between the control signal and the vehicle orientation (which will be the error determined by

ΕΡ 3 076 906 Β1 comparator), and tries to minimize it. By default, the feedback (not shown) is pre-tuned (experimentally based), but the response time is modified adaptively according to the error / difference value by matching the proportional, integral and derivative constants (used as fine tuning parameters) in narrow, experimentally determined ranges. For higher difference values, more aggressive system intervention is used to minimize error, while for smaller error values, the system responds with smaller scale interventions. The intervention is performed by the Drift Compensation Unit which, in the case of a wheelchair, adjusts the left and right wheel speeds by the appropriate amount. This solution significantly improves driving stability in the above-mentioned difficult cases, i.e. not only when driving straight.

[0048] Figs. 10A and 10B illustrate some (preferably software-implemented) modules of preferred embodiments of the control device according to the invention, and also show a block diagram of the operation of the control device.

[0049] By means of the initiating module 40 of the control device, the wheelchair 10 is configured by user actions, preferably prior to operation commencement. As a first step, the motion detection units 11, 12 built according to Fig. 2, or more precisely, their orientation values ​​are set as a base state with respect to a common frame of reference. As a second step, the user-specific peaks are entered into the peak value module 42. To do this, the user has to tilt the head sharply back and forth and turn it left and right.

[0050] In addition to the peak module 42, the processing unit 13 comprises a quartion vector computation module 41 adapted to compute the relative orientation by using preferably the formulas shown in Fig. 10A. The normalized orientation value is obtained at the output of the normalization module 43. For the subtraction operation, i.e. In releasing the signal from the sensor attached to the body from external disturbances to which the wheelchair 10 is subjected, the quaternion based method best for calculating the relative orientation is used making the operation matrix redundant. The subtraction operation provides a control signal from the user to the wheelchair 10.

[0051] The preferred embodiment shown in Fig. 10A further comprises a gesture recognition module 44, which is preferably implemented on a smartphone and uses a neural network. The gesture recognition functionality allows additional commands to be input, for example to enter a fixed target speed of the wheelchair 10, to set speed levels or to stop the wheelchair immediately. By recognizing the specific gestures or the gestures learned by the control device, the commands associated with the certain gestures are performed instead of generating a control signal from signals from the motion detection units 11, 12 as described above. The preferred embodiment comprises a gesture recognizer 44 adapted to recognize the gestures performed using the first motion detection unit 11, and when a certain gesture is recognized, the sending module 35 sends a control signal corresponding to the command associated with the gesture instead of sending an inertial control signal. In this way, the gestures can be separated from the control signal.

[0052] Smartphones with higher processing power are able to use the variable window size sampling technique to recognize the movements of the headset on the head, i.e. the temporal reorientation of the head. To achieve this, spatial surface boundaries are generated, which can be quickly and efficiently tested using their local characteristics and can be recognized in a manner consistent with an allowable variation.

[0053] When smartphone applications are used, also unintentional high-frequency driver convulsions and seizures that occur from time to time can be filtered out by gesture recognition. The maneuvering dynamics can be learned simultaneously by the driver and the vehicle. The self-learning application can be manually switched to intensive learning mode in which the gesture recognizer 44 remembers the spatial envelope of the sequence of movements. As described above, these movement patterns can be interpreted as commands and can be assigned to any functions such as: emergency stop, call answer / reject, start external applications. For example, if the sensor moves, the resulting change

ΕΡ 3 076 906 Β1 can be communicated to the control device - for example by gesture control or pressing a pressure sensor 30 provided on the head assembly; The control device calculates the deviation and the compensated test signal of the vehicle. In the event that the first motion detection unit 11 falls or the vehicle overturns, the vehicle control may be automatically locked and an emergency call will be made.

As shown in Fig. 10B, the main branch of the block diagram includes the step of inserting an inertial control signal generated from a normalized difference signal (relative orientation) into a Kalman filter 45 that performs a predicted comparison of the current control signal value with the previous value in a known manner, filtering out thus, unrealistically large changes resulting, for example, from convulsions. The Kalman filter 45 is adapted, i.e. is tuned to the style of movements of the user in question, preferably by adjusting the Kalman gain in a manner known per se.

[0055] A further adaptation is made by parameterizing the dispatch module 35 adapted to receive the filtered inertial monitoring signal. The dispatch module 35 receives as input whether the user is controlling the wheelchair by tilting or turning the head. Preferably, an internal or external dead space is also defined. The inner dead space serves to prevent the wheelchair 10 from moving violently, as small head movements will not start the wheelchair. Outer dead space is selected on the basis of the "normal" amplitude of body movement, which means that turning or tilting the head within an amplitude within the outer dead space does not normally occur. Thus, a signal from the outer dead space may, for example, indicate that the user is unconscious. The shipping module 35 thus only affects the movement of the cart in the event that the filtered inertial control signal is from outside the dead space (s).

[0056] By switching the input of the shipping module 35, the user can switch from using a headphone to control a vehicle to using a mouse with a tablet attached and vice versa. The user can switch between the two modes by, for example, pressing the pressure sensor 30.

[0057] The smoothing module 46, which receives the output signal from the shipping module 35, preferably consists of components known per se which are shown in the figure; it makes the inertial control signal after filtering and applying dead spaces to it, also smoothed out so that it does not contain high acceleration values ​​and sudden changes of direction. The current motor control signal is obtained from the output of a smoothing module 46 that performs an adaptive ramp that smoothes the signal.

[0058] Fig. 11 shows the graphics displayed on the exemplary user interface required to fit the headset to the head, and in particular the first motion detection unit 11 attached thereto. Right, left, forward and backward directions can be changed by pressing the buttons indicated by the arrows. as well as forward / right. The first element of the next row of buttons is used to turn the stabilization provided by the platform on and off, the "0" button is to set the wheelchair's orientation with respect to the direction of travel, and the square button can be used to calibrate the head set as well as to choose between tilting the head in. side and head rotation as steering actions for turning the wheelchair.

[0059] The first slider below the row of buttons is used to set the initial signal smoothing factor, while the slider below is used to set the dynamics of the control signal.

[0060] The outer concentric circle on the right shows speed, while the inner circle shows rotation. The dashed circles represent the end state for which the user cannot turn or tilt the head, and the circles without contours indicate the boundaries of the neutral zone, which can also be interpreted as joystick dead space, which means that movements inside this zone do not affect the motors.

[0061] The invention can also use other features of smartphones, for example communication via the Internet, and the body-mounted sensor, i.e. the first motion detection unit 11, can also be used to move the mouse cursor. Other similar settings, mouse settings can also be obtained via the communication interface. All settings, including adjusting the setting system, can be advantageously done in two ways: by using the touch panel of the mobile device

ΕΡ 3 076 906 Β1 or by moving the cursor using the sensor attached to the body. To click the mouse, the user can choose from four different modes: in the first mode, the mouse is automatically clicked when the cursor enters an area of ​​the selected size for a given time, in the second mode, the mouse is activated by sound recognition, for example by clicking with the tongue, saying a word or sound of a certain frequency. The third mode is related to face recognition, while the most preferred is the fourth mode, in which mouse clicks can be performed by blinking the left or right eye respectively. By using the WIFI or Bluetooth connection of the smartphone, the functionality of the mouse can be extended to support other computers. Using the cursor and the virtual on-screen keyboard, speech impaired users can enter text that is read aloud by the software's speech synthesizer module. The speed and tone of the speech from the synthesizer can be modified using the same interface.

[0062] The control device according to the invention has many advantages over existing solutions. In the present invention, the integrated sensor data describing the movement of the vehicle and rider are treated as separate inputs. A significant advantage over solutions employing a single sensor unit is that signals interfering with vehicle motion control can be filtered out with high certainty without reducing vehicle response time. Unevenness of the surface, descents and ascents do not affect the vehicle control, and you can use the full power of your vehicle throughout the speed range. Using self-learning software, it is easy to intuitively drive the vehicle, which becomes safer. Using the system offers countless additional benefits related to internet communication and the operation of electronic devices and computers.

[0063] An important advantage of the invention is that its operation is based on the processing of signals from two - preferably identical - sensor arrays: one relating the state of the body part moved by the user to the external reference system and the other relating the vehicle state to the same reference system, and by generating a vehicle control signal from relatively measured relative signals, making signals from a body-mounted sensor, and by is a control signal of the vehicle, independent of external sources of interference.

[0064] Another significant advantage is that the sensor array may consist of microelectromechanical sensors, i.e. a three-axis accelerometer, a three- axis gyro and a three-axis magnetometer. In this way, the sensor units may be used to determine the angular position and rotation of an object relative to the Earth's magnetic and gravity field. By using the magnetometer as a compass, you can compensate for gyro drift by merging sensors. As the sensors used are extremely reliable mass products, they are embedded in most known information technology products such as smartphones, tablets, cameras where they are mainly used for navigation and automatic screen rotation.

[0065] Accordingly, built-in tablet sensors can be used to measure vehicle position / orientation as the tablet is also capable of displaying user interfaces and configurations. In addition, the internet-connected device provides a wider range of communication and navigation applications for people paralyzed below the neck, who can extend their activities to using videophones, global positioning, voice searching the web, on-screen keyboard and speech synthesizers.

[0066] Another significant advantage is that although the measurement of the orientation / position of a body part requires a special device consisting of the above-mentioned sensors, a microprocessor responsible for integrating the sensors and a radio transceiver for data transmission, the software is capable, in addition to receiving signals from a special device, to receive signals from other position sensors that communicate via Bluetooth, WIFI or USB. Thus, other commercially available devices such as (built-in sensors) smartphones, joysticks, game consoles (e.g. Nintendo Wll, Xbox controller) and 3D mice (e.g. Space-Mouse) can also be connected and used for the purposes of the invention. .

[0067] Another advantage of the invention is that vehicle control software can be developed with the use of smartphones in mind. This is important as people caring for disabled people often use secondary joysticks to steer the wheelchair when the wheelchair owner is

ΕΡ 3 076 906 Β1 temporarily unable to drive or not in the wheelchair. By running wheelchair control applications on their phone, the caregiver can use the phone or a joystick to steer the wheelchair.

[0068] A further advantage is that, thanks to its modularity, the invention allows the connection of several different devices such as a joystick or a mouse, while maintaining the driving stability achieved by inertial sensors. Modularity appears in both the hardware and software configurations of the invention, since the software components also can be selected and adapted upon request.

[0069] The control device according to the invention or the wheelchair control platform, also operated by inertial sensors attached to the body, allows signals from body-mounted sensors worn by the user or signals from movable input devices to be independent of external phenomena. by filtering out disturbances resulting from uneven road surfaces, thus ensuring that the vehicle speed is kept constant on both uphills and descents, compensating for the drift of the vehicle from the direction of travel without user intervention. The digital filters contained in the wheelchair controller can be parameterized to suit the needs and capabilities of the user, allowing for adaptive control of the vehicle movement.

[0070] The platform of the invention allows persons who are to control a wheelchair to connect various input devices to the control electronics of the wheelchair in order to ensure safe driving of the vehicle.

[0071] The invention is of course not limited to the preferred embodiments described in detail above; further modifications, combinations of features and variants are possible within the scope of protection as defined in the claims.