GB2465020A - Self-balancing single-track electric vehicle - Google Patents

Abstract

An electrically propelled vehicle with two or more in-line wheels 11b, 11b is provided, one or more of the wheels 11a, 11b is rotatable about a vertical axis to allow steering of the vehicle. Electric traction motors 12a, 12b are coupled to the wheels and capable of effecting forward and reverse motion and the extraction of kinetic energy upon deceleration of the vehicle. A balancing mechanism under an automatic control system which is optionally a discrete-time controller is provided which uses sensors, for example lateral accelerometer 27, to estimate the roll moment of the vehicle in real time and adjust the center of mass of the vehicle laterally, preferably by moving the vehicle batteries 15a, 15b laterally via electric motors 16a — 16d. The balancing mechanism is capable of imparting dynamic balancing forces to the vehicle optionally by a pair of gyroscopes in assembly 14 which are rotated about their precession axes to produce, thereby keeping the vehicle upright or in any other desired orientation.

Description

Self-balancing single-track electric vehicle This invention relates to battery-powered single-track vehicles, specifically vehicles in which passengers are not seated astride the machine, as might be the case with a motorcycle, but seated inside the vehicle.

Electric vehicles using batteries as the power source are well-known. The various environmental and performance benefits of battery-electric vehicles are well documented and need not be repeated in detail.

To summarize: electric vehicles may use energy from a wide range of energy sources such as solar, wind, or wave power which are otherwise unsuitable for on-demand use, and may act as distributed storage for such power sources; when ultimately powered from conventional sources, polluting emissions are centralized at the generating plant and therefore more easily scrubbed; electric motors have a flat torque curve and do not need variable transmission, and they may be overdriven for short periods, such that the size may be smaller and the rated power output significantly lower than a heat engine of equivalent performance; electric drives and motors are highly efficient, even after taking account of power transmission and storage losses, and can operate in a regenerative mode to return braking energy to the battery. However, prior art simply addresses the mundane problem of replacing internal combustion engines in passenger vehicles with an equivalent electric motor, without considering that an electric drive system might enable improvements to the vehicle itself. Most seriously, the large battery mass has always been considered "dead weight"; in the case of existing designs, it is exactly that, occupying a great deal of space while performing no function except storage of energy. Furthermore, because no attention is paid to the form of the vehicle, the battery mass is far larger than it needs to be.

Vehicles in general typically have either three wheels in a triangular formation, or two or more parallel sets of wheels, so that the vehicle is statically stable. Single-track vehicles such as cycles and motorcycles are not statically stable, so the rider sets his feet on the ground when the vehicle is stationary, or deploys a kickstand. Dynamic balance is achieved, while the vehicle is in motion, by a combination of shifting body weight and steering action.

Two-wheeled single-track vehicles which remain upright without help from the user are not so well-known, although there are examples of prior art which provide an active control system and actuators to balance a vehicle on tandem wheels. Typically, such designs address the problem of balancing monorail vehicles, using one of two techniques: lateral movement of a large mass to alter the vehicle's centre of gravity, or modification of a massive gyroscope's natural precession response to tilting. US Patent No. 3124007 is an example of the former, while GB Patent No. 26034 is an example of the latter.

In order to appreciate the problems with existing self-balancing vehicles, it is important to review why such a vehicle, in general terms, might offer advantages compared to a multi-wheeled system. The improvements over prior art offered by the present invention may then be more obviously apparent.

We consider first the issue of stability and vehicle handling. A vehicle with parallel pairs of wheels is statically stable, but its dynamic performance at speed is extremely poor. Complex techniques must be used to constrain and control the relative movement of the wheels on uneven surfaces or when turning corners.

To date, no single optimum solution to this problem has been found, with all prior art involving some suboptimal tradeoff in performance, complexity, or cost.

For three-wheeled vehicles, dynamic stability tends to be even worse: they are prone to tipping on corners, and proposed solutions to this problem -such as a tilting chassis -are generally even more complex than for four-wheeled vehicles.

Contrasting with the above, a two-wheeled single-track vehicle is statically unstable, but has far superior dynamic performance with a much simpler mechanical configuration. An effective two-wheeled road vehicle can be implemented with very few steering and suspension components, as exemplified by cycles and motorcycles. Unfortunately, cycles and motorcycles offer no protection to the rider, who is exposed to environmental discomfort and danger from other vehicles, hence the desirability of an enclosed chassis and a self-balancing design. In the following discussions, the term single-track vehicle' refers to such a self-balancing vehicle, and specifically excludes cycle-like machines.

Both road and rail vehicles with a parallel-wheel configuration are prone to catastrophic tipping on corners.

Cars are made wide and flat, with complex wheel geometry control, to ensure that tipping is unlikely to occur; rail vehicles have enforced constraints on speed and turning radius; roads and railroads must be constructed with appropriate camber. Unless wheel adhesion is completely lost, a two-wheeled vehicle simply leans to balance cornering forces.

In the particular case of rail vehicles, a particular dynamic instability known as hunting' occurs with dual-track vehicles running on an adhesion railway. This is an oscillatory behaviour which usually occurs at higher speeds, whereby the vehicle sways laterally and may topple or derail. This problem can be solved by precise engineering or with active control systems, or a combination of the two. However, a vehicle with in-line wheels on a single rail is inherently immune to hunting oscillation.

Considering now the issue of aerodynamic efficiency, the typical four-wheel layout of a passenger vehicle presents several problems. As described above, the vehicle must have widely-spaced wheels to guarantee stability. Since it must also be high enough to accommodate human occupants, a minimum limit on cross-sectional area is imposed, about 2 square metres. The front area of a typical small passenger road vehicle is 2.5 square metres. Furthermore, the exposed wheels and wheel recesses contribute to a higher drag coefficient. These two factors result in a high rate of energy loss due to aerodynamic drag at higher speeds, typically 10-20kw at lOOkph, depending on the vehicle design.

A two-wheeled vehicle need only be as wide and as high as a seated human, about 1 square metre. The wheels are disposed along the centreline of the vehicle, protruding from an otherwise unbroken underside surface, arid contribute little to the drag coefficient. To seat a useful number of passengers, the vehicle may have an elongated, teardrop-like profile. These combined factors result in extremely low aerodynamic losses: typically 50-70% less than a four-wheeled vehicle of equivalent passenger capacity.

This improvement permits a smaller engine, or equivalently, a smaller motor and battery. This affords a further weight reduction, resulting in a vehicle significantly lighter than a traditional car. This in turn results in a reduction in rolling resistance. Rolling resistance can be reduced further by using more advanced tyres with a lower rolling resistance coefficient -economically a more practical proposition with two wheels instead of four. Whereas rolling resistance accounts for 8-12KW of engine output in a typical four-wheeled car moving at lUUkph, a two-wheeled car may reduce that by 40-50%.

Despite the theoretical performance advantages of an enclosed, self-balancing single-track vehicle, there are serious problems with existing designs and such vehicles have not seen widespread adoption.

we review first those vehicles which use the precession of a massive gyroscope to supply a righting torque under the influence of a tipping moment. According to this design, the fundamental property used for balancing is precession: a torque applied to the gyroscope about any given axis will result in a rotation at right angles to the original torque and to the spin axis. Thus, provided within the vehicle are one or more gyroscopes which can precess freely, arranged such that either the precession axis or the spin axis is vertical.

Consider such a gyroscope mounted in a vehicle with the plane of the flywheel perpendicular to the plane of the vehicle floor and parallel to its line of axial symmetry -that is, with the spin axis horizontal and across the body of the vehicle. The gyroscope should be supported such that the flywheel may pivot freely about an axis perpendicular to the floor of the vehicle. If the vehicle experiences an unbalancing force and this force is transmitted through the vehicle body as a torque about the vehicle's roll axis, the gyroscope will simply rotate around the precession axis -the perpendicular -effectively preventing the unbalancing force from toppling the vehicle. However, if the unbalancing force is not removed, the gyro will continue to precess until its spin axis becomes parallel to the vehicle's longitudinal axis. Since the precession axis rotates with the spin axis, when this point is reached the precession torque is no longer about the roll axis but about the pitch axis, and the gyro can no longer balance the vehicle.

Any practical gyroscopically balanced system must therefore incorporate an active control system. The function of this system is partly to cancel the friction which resists the gyroscopic precession, and partly to re-inject the energy lost as the vehicle tips, thereby preventing the gyro from precessing to the point where it provides no restoring torque. In the literature, this is usually referred to as "accelerating the precession", since the servo acts in the same sense as the natural precession movement.

The advantage of this system is that it is inherently safe: on servo failure, the gyro will continue the keep the vehicle balanced, at least for a short while, while the operator has an opportunity to stop safely.

One disadvantage is the difficulty of implementing a reliable control system. All known prior art uses a continuous-time control system to modify the precession of the gyroscopes, which in practice has proven extremely difficult to achieve. The difficulty arises partly because the gyroscope is effectively used as both sensor and actuator: as a sensor, to convey to the control system that the vehicle is experiencing a tilting force, as indicated by a precession movement; and as an actuator, by applying a righting moment to the tilting vehicle under the control of the servo system. This problem is exacerbated by the fact that actuation operates in the same sense as the natural precession movement. Systems which exhibit this kind of response are generally difficult, if not impossible, to control reliably. The present invention solves this particular problem by using the gyroscope as an actuator only, and provides separate sensing means to determine vehicle roll rate.

Another serious disadvantage of a gyroscopic stabilizer is that it can only cancel transient unbalancing forces.

If a constant offset is present -for example, if the vehicle is asymmetrically loaded or experiences a sidewind -then the control loop will cause the vehicle to list to the left or to the right in the steady state, such that its centre of mass is disposed vertically. This may be disconcerting for the occupants.

One novel proposal for solving the issue of constant forces (US Pat. No 3124007) describes moving the gyroscope mass itself laterally within the vehicle. Unfortunately, this is unworkable unless the mass of the flywheel is a significant fraction of the fully-laden vehicle mass. In addition, the mechanism upon which the flywheel is moved must be designed to carry substantial gyroscopic forces, adding to the weight and complexity of the vehicle. Furthermore, the presence of an extremely large, heavy flywheel represents a serious hazard to passengers in the event of an otherwise survivable vehicle collision.

It is therefore the primary object of this invention to provide a single-track vehicle which will remain balanced substantially upright, whether in motion or stationary, regardless of offset loads; and which will lean at an appropriate angle while cornering.

It is a further object of the invention to provide a system which is inherently safe under conditions of mechanical failure, at least to the degree that a conventional dual-track vehicle is safe under equivalent conditions.

It is an additional object of the invention to provide a system in which the various major subsystems perform two or more functions, such that overall vehicle mass is less than, or no greater than the vehicles described in prior art while offering significantly enhanced performance.

Accordingly, the present invention proposes a vehicle with two or more in-line wheels, at least one of which is used to steer the vehicle and at least one of which is driven by an electric motor, provided with a balancing mechanism mounted rigidly to the vehicle chassis under the control of a discrete-time controller.

The balancing mechanism comprises two main subsystems, the first subsystem being the main vehicle battery mounted on a motorized sled, which may be moved laterally within the vehicle chassis under the command of the controller, and the second subsystem consisting of a pair of large gyroscopes with their spin axes in a horizontal plane and precession axes vertical, both gyroscopes being rotatably mounted in gimbals such that they may be rotated by the controller about their precession axes by means of gimbal control motors. Control inputs for balancing are obtained from an accelerometer which measures the attitude of and forces acting upon the vehicle, from sensors measuring the natural precession motion of the gyroscopes, and from sensors measuring the vehicle steering rate. Control outputs for balancing consist of a position command for the battery sled, and measured pulses of energy delivered to the gimbal control motors. The two gyroscope flywheels also serve as temporary energy storage for regenerated electrical energy produced by the traction motor or motors during braking, energy regeneration being also managed by the balancing controller.

The accompanying drawings form part of this specification and will be referenced herein.

Fig. 1 is a schematic view of the vehicle chassis showing vital subsystems. Omitted from this figure are the usual practical requirements for friction braking, steering, suspension, chassis stiffness, safety, and so on, since these are irrelevant to the present invention; also omitted, for clarity, are electrical connections.

Fig. 2 is a schematic view of the electrical interconnections between the systems shown in Fig. 1.

Fig. 3 is an exploded view of a single gyroscope, in one particular embodiment.

Fig. 4 shows two of the gyroscopes of Fig. 3 incorporated into a frame.

Fig. 5 shows the assembly of Fig. 4 in section view so that internals and drive motors may be seen.

Fig. 6 details one battery pack and its sled mounting.

Referring first to Fig. 1, a vehicle is provided with a chassis 10 upon which are mounted two or more driven wheels ha and hlb, so that the vehicle may be propelled along a road, rail, or other substrate. Said wheels are disposed along the length of the vehicle, at least one of which being rotatably mounted about an approximately vertical axis in order to steer the vehicle.

It should be noted that although Fig.1 depicts a vehicle symmetrical about the longitudinal centreline, this is not essential. In certain types of vehicle, it may be advantageous to offset the wheels or other components towards one side or the other and to adjust the mass distribution of the vehicle accordingly; for example, a rail vehicle designed to run on one of a pair of standard-gauge rails may be offset substantially to one side, so that a bidirectional rail service can be operated on a single pair of rails. It may also be desirable to include more than two wheels, whether driven or freely rotating, to maintain chassis stiffness without excessive use of structural material. The steering configuration may also be adjusted to suit the vehicle type; thus a road vehicle will most likely be implemented with a single steered front wheel with a rake and trail offset for stable performance, while a small rail car may require steerable wheels front and rear with a non-steerable wheel near the centre.

In the preferred embodiment, wheels ha and llb are mechanically coupled directly to, or form a single assembly with, traction motors 12a and 12b; the traction motors 12a and 12b are electrically connected to inverters ISa and 19b; and the front wheel is a steered wheel.

A gyroscope assembly 14 is mounted within the chassis. In Fig. 1, the casing is removed from one gyroscope so that motor 25 and flywheel 26 are visible, both gyroscopes being of identical construction. Assembly 14, as detailed in following paragraphs, performs two functions: namely, dynamic balancing of the vehicle and storage of braking energy.

Two battery packs iSa and 15b are mounted on sleds, which are detailed further in Fig. 5, such that each pack may move laterally from one side of the vehicle to the other. The lateral movement of the battery packs is accomplished by motors 16a,16b,16c and 16d, the rotary motion of which is converted to linear motion by means of leadscrew and nut assemblies 17a and 17b. The movable battery mass, which by the nature of chemical batteries may amount to 20% or more of the vehicle's overall mass, allows the vehicle's centre of mass to be held vertically above the wheels, or thereabouts.

Disposed around the vehicle are: inverter 18, which controls the rotation of the motors within gyroscope assembly 14; buck/boost converter 20, which manages charge flowing into and out of the battery packs iSa and lSb; and inverter 21, which manages power flowing into and out of the external charging supply 29, when connected. Also provided is a lateral accelerometer 27 and, in the preferred embodiment, a large braking resistor 22. A control system 23 preferably contains its own auxiliary power supply 24.

In a practical implementation, a small side wheel or kickstand is desirable, fitted on one side of the vehicle and perhaps deployable or retractable on command of the operator, so that the vehicle can be powered down completely when not in use.

Referring now to Fig.2, which shows a schematic view of the electrical connections omitted from Fig. 1, the inverters 18, 19a, 19b, and 21 are shown connected to a common DC bus 30, as is buck/boost converter 20.

A capacitor bank 28 is also attached to the DC bus for temporary energy storage. Power flow between the various elements is supervised by control system 23, which also manages vehicle balance with reference to accelerometer 27. Preferable, but not essential, is an auxiliary low-voltage battery 24, provided to maintain minimal system operation in case of failure of the main batteries.

Without changing the fundamental features of the invention, a heat engine and generator set may also be added to the vehicle, to deliver reserve power onto the DC bus when the batteries become empty or are unable to deliver the power level required for high speed operation.

Although in the preferred embodiment the inverters and buck/boost converter are physically separate and interconnected by power and control cables, there is no reason in principle why they may not be combined into some smaller number of enclosures or a single assembly.

Battery packs iSa and 15b have a nominal voltage lower than the nominal DC bus voltage. In one embodiment, the battery voltage may vary between 168V and 32OVDC, depending on state of charge and electrical load, and may require a charging voltage of up to 35OVDC; the DC bus voltage is typically maintained between 350V and 39OVDC. Thus, when power is demanded from the battery packs, the converter 20 is used to boost the battery voltage to the operating voltage of the DC bus. For battery charging, the converter operates in buck mode to regulate the current flowing from the DC bus into the battery packs.

The inverters, by their nature, permit power flow either into or from their connected motors. Inverters 19a and 19b, connected to traction motors 12a and 12b, normally deliver power to their respective motors.

Generally, power is delivered to the motors, and therefore to the wheels to propel the vehicle; if more than one wheel is driven, their relative rotational speed is maintained close to zero by their respective inverters.

During braking, the inverter operates instead in inverting mode, returning power to the DC bus -that is, to the capacitors 28-which causes the DC bus voltage to rise, and the energy must be rapidly removed if damage is to be avoided. This may happen in one of three ways, as follows.

Firstly, under light braking, the battery pack can absorb all of the energy delivered into the DC bus.

Secondly, if a large power flow occurs, as when braking rapidly from high speed, this cannot be safely or efficiently absorbed by the batteries. In a traditional electric vehicle, this problem is solved by installing a large and costly capacitor array and an associated charge management system. In the present invention, excess energy is diverted, under the control of the control system 23 via inverter 18, into the gyroscope flywheels. In the preferred implementation, the gyroscope motors are sized to accept energy for short periods at approximately the same maximum rate as the battery pack, and the rotors are sized such that accepting 50% of the regenerated braking energy from a full-speed halt increases the flywheel rotor speed by a relatively small amount, of the order of 25%. The kinetic energy thus imparted to the flywheels will remain for minutes or hours before being completely dissipated by friction, during which time it may be re-injected via inverter 18 onto the DC bus for traction motor acceleration, or for a slow battery recharge if battery capacity exists. During these operations, the balancing mechanism described below must take account of the changing gyro rate when computing balance corrections.

Thirdly, under extreme conditions, such as an emergency stop from full speed, or in the case of failure of some part of the control electronics, or in the case where neither the batteries nor the flywheels can accept more energy, the electrical energy regenerated into the DC bus is dumped into braking resistor 22, which will become extremely hot in consequence. In the preferred embodiment, the braking resistor has a planar form factor with a large surface area exposed to air beneath the vehicle. It may alternatively be incorporated into the climate control system of the vehicle to deliver heat either directly into the cabin, or into a thermal storage medium, or to an absorption refrigeration means for cabin air-conditioning.

It should be understood that flywheel energy storage is provided primarily to address limitations of the present state of the art: namely, that chemical batteries can absorb energy at only a certain maximum rate, typically at 2-5 times their rated discharge current, with decreasing efficiency and reduced cell lifetime at the higher charge rates. It might be assumed therefore that future advances in cell technology may render the energy-storage capability of the flywheels somewhat superfluous. However, this is not the case: any cell will have a maximum energy storage capacity, so braking energy can still be regenerated into the flywheels when the cells are completely full; furthermore, the flywheels will continually dissipate energy through friction, and braking regeneration represents a straightforward way to maintain flywheel speed without imposing a continual drain on the main batteries.

Referring now to Fig.3, a single gyroscope is depicted within a vacuum housing. A shaft 32 carries rotor 31, said rotor being free to rotate on bearings 33a and 33b which are pressed into the rotor body, bearing 33b being on the hidden side of rotor 31. Also mounted rigidly to the rotor body on the same axis as shaft 32 is an assembly of permanent magnets disposed in a circular fashion around an iron core 34, which, together with stator windings 35, comprise a synchronous motor, commonly known as a brushless DC motor. A bracket 36 is provided, which holds the motor stator and shaft fixed to the outer housing 37. Bracket 36 is mechanically inadequate to carry the full precession forces of the gyroscope, but is provided so that the flywheel can be spun to high speed and balanced during production, before cover 38 is fitted, and also to restrain the heavy flywheel assembly from lateral movement so that cover 38 may be fitted easily. An 0-ring gasket 39 seals the two halves of the enclosure 37 and 38, allowing a rough vacuum within the enclosure. When assembled, shaft 32 is held securely in place by the two halves of the enclosure.

In an alternative embodiment, the flywheel is spun instead by means of a disc carrying permanent magnets fixed securely to and coaxial with the flywheel, stator coils being disposed circularly around the two faces of the disc such that the magnetic circuit passes through the disc and around its edge, forming an arrangement commonly referred to as an axial flux permanent magnet motor.

In the preferred embodiment, the rotor 31 is a high-density material such as cast iron, while vacuum housing 37/38 is a stiff, lightweight material such as aluminium alloy. Bearings 33a and 33b are constructed using solid lubricants such as molybdenum disulphide or diamond-like carbon, preferably in conjunction with a heavy grease lubricant. Although such bearings operate reasonably well in a vacuum they are not suitable for the extremely high speeds normally associated with high energy flywheels; in any case, the gyro's ability to balance the vehicle depends on its moment of inertia, which is directly proportional to angular velocity, and not on its rotational energy, which is proportional to the square of angular velocity.

Furthermore, gyro nutation, the characteristic frequency of which depends on the rotor proportions, will cause significant vibration at certain rotor speeds. Considering all of these factors, the rotor mass in the preferred embodiment is relatively high, the rotational speed is relatively low, and the level of vacuum maintained in the enclosure is a modest fraction of atmospheric pressure.

The electrical connections to the motor stator winding 35 is preferably by means of three rotating electrical couplings known as slip rings attached to trunnions 40a and 40b. This form of coupling will be familiar to those skilled in the art, the primary purpose of its use being to avoid a mechanically-unsecured cable bringing power to and from the gyroscope internals.

It will be appreciated that prior art encompasses many possible embodiments of the basic flywheel design, which is a common engineering component. Any such design may be successfully applied to the vehicle described herein without any modification to the rest of the system. Fig. 3 merely describes one possible implementation without any intention to restrict the scope of the invention.

Referring to figure 4, the gyroscope assembly annotated as item 14 in Fig. 1 is illustrated in detail, although somewhat schematically, with two of the gyros of Fig. 4 mounted in a two-part frame, 40a and 40b, which is of a light and stiff material such as cast aluminium alloy or carbon fibre composite, and acts as a gimbal. In the preferred implementation, frame 40a/40b is fixed rigidly at points 41a, 41b to the vehicle chassis.

At one end of each gyroscope, gears 43a and 43b are rigidly attached such that the gyroscope will rotate on its trunnion when the gear is driven. Motors 44a and 44b are rigidly fixed to the frame adjacent to gears 43a and 43b, and engage with said gears via pinions 45a and 45b such that the gyroscope may be turned about a vertical axis by applying power to either one of the motors.

Referring now to Fig. 5, the same assembly is seen in section view. The gyroscope trunnions are supported by suitable bearings within the frame; in the preferred implementation, each trunnion is rotatably mounted within needle bearing assemblies 46 and supported axially by thrust bearings 47.

The gyro stator windings 35 are connected in parallel to inverter 18 of Fig. 1, such that the motors rotate at precisely the same speed; however, the phases of the two motors are connected such that one gyro motor rotates clockwise, and the other rotates anticlockwise. Additionally, Fig. 4 shows a pair of spring steel bands 48 is attached to the circumference of each driven trunnion, crossed at the midpoint between the two gyros, such that precession or driven rotation of one gyro will cause a synchronous counter-rotation of the other.

The method of two linked counter-rotating gyros ensures firstly that unwanted gyroscopic torques in the fore-aft direction do not affect the vehicle attitude, such torques being neutralized by virtue of the heavy, stiff mounting frame within which the gyros are mounted; and secondly that gyroscopic roll torque developed is of substantially the same magnitude and direction from each gyro. In an alternative implementation, the two gyros are not mechanically linked, and instead the two motors 44a and 44b are driven in synchronism such that the rotation of each gyro is equal and opposite to that of the other.

However, this does have the disadvantage of degraded performance if the servo partially or entirely fails.

Linking the gyros allows a degree of redundancy: in the case of the failure of one motor, the remaining one can still operate both gyros.

Inverter 13 is preferably a sensorless type, requiring no explicit position feedback from the rotors, and incorporates contactors and load-sensing firmware to ensure that, in the case of failure of one gyro motor, the faulty motor may be disconnected and the remaining motor may still operate normally.

In the preferred implementation, motors 44a and 44b are coreless motors since, when not energized, they must offer the smallest possible resistance to the precession of the gyroscopes. Coreless motors are typically very poor at dissipating heat from their windings; however, as will become apparent, motors 44a and 44b are operated with a very low duty cycle, mitigating this particular drawback.

The operation of the gyroscope assembly is as follows.

Control system 23 samples at intervals the attitude of the vehicle by means of lateral accelerometer 27, the absolute angular position of each gyro as it precesses on trunnions 40a and 40b, and the instantaneous velocity of precession of each gyro by monitoring the back EMF generated by motors 44a and 44b. The control system is also aware of the angular velocity of each flywheel -inverter 19 must accurately measure this to function correctly. The absolute position of each gyro may be measured directly using any of several commonly known means. In one embodiment, the position is estimated from the velocity measurement in combination with a single index hole located on gear 43a which is sensed by an inductive sensor. In another embodiment, the face of gear 43a is marked with a suitable absolute angle code such as a Gray code, and sensed bya multi-channel optical sensor.

Given this information, the control system can estimate at each time interval T how much energy the vehicle has lost, or gained, due to external disturbing forces since time T-1. If a disturbance has tipped the vehicle away from the upright position, the control system may then inject the calculated amount of energy by actuating motors 44a and 45b for a short time interval t, where t<T, while measuring or controlling the actuation power; the motors are actuated such that the gyro is driven in the same direction as its measured precession. The gyro will then be left to precess freely for the remainder of time interval T. Since the energy estimate is unlikely to be entirely accurate, at time T+1 the gyro will most likely be precessing in the same direction it was previously, except at a slower rate, or it will be moving slowly in the opposite direction.

In the former case, not enough energy was injected; in the latter, too much. Assuming the control parameters are approximately correct, and in the absence of further disturbances, the gyro precession rate will settle to zero.

In the preferred embodiment, the control system incorporates an observer process which continuously modifies the control parameters to optimize the system response.

It should be noted that the control system's target is not only to maintain the vehicle attitude but also to keep the gyros somewhere near their null position -that is, with the spin axis across the width of the vehicle body. This requires, firstly, that the vehicle centre of mass is, averaged over the long term, held approximately perpendicular to the road or track, by means of the mechanism described in the following paragraphs; and secondly, that the gyro control overcompensates slightly, so that the gyro is not just impelled to remain at a stationary, offset angle, but actually pushed back towards the null position at each sampling interval, preferably at a velocity proportional to its angular displacement. In other words, the control loop should be slightly underdamped.

It should also be noted that the data from acceleration sensor 27 will be heavily corrupted by noise, caused by vibration from various sources. In the preferred embodiment, this is mitigated somewhat by using a dual-axis accelerometer, arranged to measure both lateral and vertical acceleration, and by using appropriate suspension to attach the accelerometer to the vehicle chassis. In the preferred embodiment, the controller 23 also filters the accelerometer readings to substantially attenuate signal components above a few hertz.

It should be further noted that an alternative arrangement of gyroscopes is possible, such that the spin axes are vertical and the precession axis is across the body of the vehicle. This has the advantage of reducing vertical height of the balancing mechanism, but has the disadvantage that pitching of the vehicle -as will be normally experienced while ascending or descending hills -will cause the gyros to precess and generate a tipping moment.

We now refer to figure 6, which shows a single battery pack 70 mounted on a movable sled 71. In the preferred implementation, the vehicle contains two such battery/sled assemblies as shown in Fig. 1, although this is not a critical aspect of the invention. The battery is moved laterally by control system 23 such that the vehicle's centre of mass is disposed perpendicular to the longitudinal centreline of the vehicle's wheels. The response of the control loop is tuned to respond only to persistent changes in centre of mass, such as might occur with luggage loaded on one side of the vehicle. Transient toppling forces -such as would occur at the moment luggage is deposited in the vehicle -are cancelled by precession of the gyroscopes. As the battery sled responds, over a matter of seconds, the gyroscope control loop will be able to return the gyros to their null position.

The sled position demand signal is derived primarily from the tilt angle estimation provided by accelerometer 27, but also takes account of gyro displacement and velocity; if the gyros are rapidly approaching their limit stops -that is, with their spin axes parallel with the vehicle's longitudinal centerline -the sled position demand will be modified such that the gyros are driven quickly back towards their null position. This limit-avoidance action is unlikely to occur in normal operation if the gyros are conservatively sized.

The battery/sled assembly operates as follows. Sled 71 is driven towards the left of the vehicle by powering motor 72, which turns leadscrew 73 and therefore draws nut 74 along the length of the thread. Likewise, the sled may be driven towards the right by means of motor 75. The sled is supported on its opposite side by rod 76 and linear bearing 77.

Leadscrews are relatively inefficient, and the motors are locked by the control system when not turning, so the battery sleds will not backdrive when subjected to lateral forces, such as might be experienced during sharp cornering.

In an alternative configuration, a single motor may be used to turn the Ieadscrew, but a dual-motor arrangement provides redundancy, simplifies the control electronics, and, with correct motor control, allows for faster and smoother actuation; however, this is not a critical aspect of the invention and other arrangements are possible. In another embodiment of the invention, the Ieadscrews are powered from one end only, and the single motor is driven from a circuit which allows rotation in both directions.

In the preferred embodiment, motors 72 and 75 are geared such that the total minimum transit time of the sled, from one side of the vehicle to the other, is of the order of three seconds. This ensures that a typical centre-of-mass disturbance can be compensated completely before the gyros have rotated through more than a few degrees of arc.

Also in the preferred embodiment, at one extreme of the sled's motion, the battery protrudes from one side of the vehicle, after removal of a suitable access panel, for manual removal or maintenance.

Alternative arrangements for lateral movement are also possible. For example, the battery pack may be suspended at either side by a suitable four-bar linkage providing for straight-line motion, with the motors driving one pivot of the linkage; such linkages have the advantage that their motion can be adjusted to provide a slight upward tilt at one end of travel, thereby making service removal easier.

It should be noted that the forces applied to accelerate the battery pack laterally will produce equal and opposite disturbing forces on the vehicle body which will cause the gyros to react. The control loop must be carefully tuned to prevent a heavily underdamped response -wobbling of the vehicle -due to this effect.

Because of the complexity of the multi-input, multi-output control loop, and the variable characteristics of the vehicle itself, the control system architecture used in the preferred embodiment is a self-tuning fuzzy controller. Starting from a reasonable approximation of the vehicle dynamics and accounting for the fundamentally nonlinear behaviour of the gyros, such a system is able to rapidly evolve optimal control parameters.

One benefit of the proposed control architecture is that the vehicle can lean naturally on cornering. An ordinary cycle or motorcycle requires the user to countersteer on entering a corner to cause the vehicle to lean into the upcoming turn; the present invention, however, is able to initiate a lean by actuating the gyros.

Accordingly, the preferred embodiment provides for the control system to initiate a lean proportional to steering wheel angular acceleration, forward speed, and estimated vehicle mass, subject to calculated safety limits. As the vehicle enters the turn, the control system naturally maintains the correct lean angle by attempting to null the centrifugal force mv2/r, as measured by the accelerometer 27. Furthermore, if the control system estimates the vehicle's forward speed is excessive for the demanded rate of turn, the speed can be automatically reduced before the lean and turn occurs.

In a similar fashion, the gyros may be used to lift the vehicle upright in the event of a complete capsize, perhaps due to an accident, or if the vehicle has been intentionally set to rest on its side, perhaps for repairs.

This can be achieved even with a relatively small gyro, as follows. The gyros are first rotated to one limit stop, then spun to full speed. The gyros are then rotated rapidly under closed-loop control -typically, all the way around to the opposite limit stop -thus lifting the vehicle into the upright position.

A further benefit is the extremely robust performance of the control system. The control architecture proposed herein differs fundamentally from the methods proposed in prior art -typically continuous-time implementations which attempt to accelerate the precession' of the gyroscopes. Although such methods have been proved marginally workable, it is extremely difficult to guarantee the stability and robustness of the necessary positive feedback loop except under a narrow range of conditions. The intermittently-acting, self-tuning implementation used in the present invention is not only more robust, but uses the actuator motors much less aggressively, ensuring greater reliability and somewhat lower power consumption.

A further benefit is that the vehicle will remain upright even with multiple system failures, at least for long enough to bring the vehicle safely to a halt. If the battery sled mechanism should fail, the gyro subsystem will be able to maintain the vehicle balanced indefinitely, although not necessarily level. If the gyro system should fail, the vehicle can still use the battery sled subsystem to keep the vehicle approximately level. If the accelerometer or other sensor should fail, the control system can fall back to a simpler control mode, using information from the remaining sensors to maintain vehicle attitude, although the vehicle will eventually begin to heel over to one side or the other. In the unlikely event of a complete system failure, including failure of the auxiliary power source, the battery sled will most likely already be in the correct position for level running, and the gyros will continue to precess freely until they spin down, thereby preventing the vehicle from toppling for several minutes.

Yet another benefit is that the heavy battery, often cited as a major drawback of electric vehicles, is put to good use by maintaining the vehicle in a level orientation regardless of offset loading. Single-track vehicles which do not have such a large mass available must, of necessity, operate with the loaded side tilted upwards in order to balance at all.

An additional benefit is that the gyroscopes, which otherwise might also appear to be excess weight, performs double duty as part of the energy regeneration system. Vehicles are known in prior art which use a flywheel as primary energy storage, but, as with a chemical battery, the flywheel in such vehicles is essentially dead weight, and the associated gyroscopic effects are considered a nuisance that must be managed rather than a useful effect to be exploited.

It will be understood that the foregoing description, while making primary reference to road and rail vehicles, is for illustration only and the essential components may be applied to any form of electrically-propelled single-track vehicle, whether for passengers or for freight; thus the embodiments described herein do not represent any specific restriction on the scope of the invention.