GB2563228A - Propulsion frame - Google Patents

Abstract

A hovering air vehicle (HAV) 200, such as a quad-rotor HAV, has a propulsion frame which is mounted to a main body (or pod) 210 via a compliant interface. The propulsion frame is made up of a plurality of arms 202 connected to a central frame portion 208 and with motors 204 and propellers 206 mounted to their distal ends. An Inertial Measurement Unit (IMU) 220, provided for flight control of the HAV, is mounted within a carrier 222 which is itself mounted to the central frame portion 208 of the propulsion frame by carrier springs 224 forming a further compliant interface. The compliant interface between propulsion frame and main body comprises soft dampers in the form of helical springs 212 and motion limiters 214 made up of stops 216 enclosed within limiting retainers 218.

Description

Propulsion Frame

FIELD OF THE INVENTION

The present invention relates to a propulsion frame mounting arrangement for a compact hovering air vehicle (HAV) equipped with sensing devices which are sensitive to vibration.

BACKGROUND

Small, hovering air vehicles (HAVs) that are portable, capture high quality imagery and can navigate without GNSS (Global Navigation Satellite Systems such as GPS, GLONASS, Galileo or Beidou) are required for many applications involving the need to quickly capture imagery using an aerial camera platform.

The airframes of most HAVs have multiple propulsion units set off from a centra! pod. In the case of a quad-rotor, which has four propulsion units, the typical airframe combines four propulsion units and a central pod. There are four arms that connect the four propulsion units to the central pod, and these arms can be fixed, or they can be movable between a deployed position and a stowed position. An airframe with fixed arms is relatively large in size, and not compact enough to be easily carried, whereas an airframe with movable arms permits the airframe to be folded into a smaller, stowed volume, allowing for easier carriage. There are many designs of HAV airframes with different arrangements of movable arms. Small HAVs, such as quad rotors, are typically less than 250g in weight and fold into a smaller volume for enhanced portability, to be stowed in a pocket or pouch, for example. A small HAV system comprises an air vehicle and a ground control station (GCS). As well as being flown in the open air, smalt HAVs are designed to be flown in cluttered, 'low-air' environments such as in urban areas below roof-top level, in forests and indoors. GNSS is the most commonly used geo-location sensing on HAVs and provides a stream of time-stamped, geo-iocations (fixes) at high rates - 10 Hz is typical. GNSS is based on very low power RF signals transmitted from satellites. GNSS has drawbacks for HAV navigation including: GNSS usually does not work indoors, GNSS can have errors from multi-path signals of the order of 40m in urban canyons and GNSS signals can be jammed or "spoofed".

Navigation of a small HAV with reference to a route on a geo-referenced map requires the small HAV system to have a geo-location system, or a dead-reckoning system, or both, GNSS, when neither jammed nor spoofed, is excellent for open air navigation of a small HAV without dead reckoning. In low-air, GNSS is not feasible as a reliable means of navigation.

An alternative HAV geo-location approach is to provide a database of geo-referenced visual landmarks - such as are being developed by many groups for autonomous navigation of cars on roads. The small HAV recognises the visual landmarks and takes accurate bearings from them as it moves. The visual landmark data is combined with other sensors and a clock to generate fixes. HAVs can be flown in low-air anywhere (not just above mapped roads); databases of HAV-relevant, geo-referenced visual landmarks, if available, are likely to be sparse. Alternative geo-location approaches to GNSS for providing good-quality fixes may improve over time.

For a mission in a low-air environment, it is currently necessary to provide a dead reckoning system either for the whole flight or for the periods between good-quality fixes. It is beneficial if the cumulative dead reckoning error is small. A CEP (circular error probable) of < 3m is of some use for navigation in less cluttered environments. A CEP of an order of size of lm is useful in cluttered urban environments. A Kalman filter such as an Extended Kalman Filter (EKF) or other data fusion method can be used to improve dead reckoning estimates (which generally have increasing CEP over time) with fixes (each fix has an associated CEP estimate). HAVs are most often flown at heights above ground level (HAGL) of up to 30m HAGL, but are also flown in the range of 30-120m HAGL and less commonly at more than 120m HAGL (120m - or 400 feet - HAGL is a common height specified by airspace regulators as the maximum permissible HAGL for a small HAV).

On manned aircraft, dead reckoning has been provided for decades using heavy, inertial-only systems and working over thousands of kilometres. There is no practicable inertial navigation sensing system available for small HAVs weighing less than 250g that, without other sensors, provides dead reckoning with a small cumulative CEP error for a flight duration of around 20-30 mins with cumulative displacement over the ground varying from tens of metres to over 2,000 metres. Practicable, available inertial navigation sensors are MEMs (micro-electro-mechanical systems) IMUs (inertial measurement units). A MEMS IMU on its own has a low CEP for just a very short time - of the order of a second (as opposed to several milliseconds or several minutes). The CEP error of a MEMS IMU grows more rapidly in a high vibration environment than in a low vibration environment.

Robust and effective, dead reckoning systems for small HAVs are still at the research stage. Several prototype dead reckoning systems for HAVs have an inertial sensing component and a visual sensing component. Other sensing components such as magnetometers and barometers are often incorporated in prototypes. In the visual sensing component, a stream of images from one or more cameras mounted on the HAV is processed to follow visual features and compute the motion of the camera(s) between frames and keyframes. For the visual odometry component to function well, it is necessary to minimise the level of vibration in the camera(s) used for dead reckoning. Data from the inertial and other sensing components is processed with data from the visual sensing component. The cumulative CEP error of the dead reckoning process is smaller when data from all the sensing components is brought together in a process that provides good dead reckoning. Processes for combining data from sensing components for stable and low-error dead reckoning in a range of real-world, noisy environments is an active area of research and development aimed at improving robustness and reducing error. A filter such as an EKF can be used in a process for bringing together components to provide good dead reckoning.

To have robust control of a HAV such as a quad rotor, it is current practice to execute a low-level control loop at high rates of typically 750 Hz or more. The low-level control loop uses at least inertial sensors and the output is the motor demands to the four motors. A key design issue is reducing the lag time between the inertial sensing and outputting changes in motor demand. Most low-level control loops use a PID (proportional, integral, derivative) control and one of the key development tasks in commissioning a low-level control loop is to set the gains, a process which is well understood by a person skilled in the art.

The vibrations in a small HAV can be excessively high if at least one of the propulsion units is out of balance compared to when all the propulsion units are in balance. In most HAV operations, care is taken by the user to ensure that all the propulsion units are in balance before a flight takes place. The majority of imbalance issues can be solved by simply replacing an out of balance propeller with a spare, balanced propeller. When a small HAV must be deployed urgently, it is an advantage if the small HAV can operate well (defined as at least: robust / responsive flight control, high quality visual imagery and sufficiently accurate dead reckoning) when one or more propulsion units is out of balance and generating excessively high-amplitude vibrations.

In general, stiff airframes have lower vibration energies than less stiff airframes. To minimise vibration, the airframes of HAVs are designed to be stiff. The provision of a stiff airframe, assuming good engineering practice in pursuit of weight optimisation, results in a heaver airframe than that of a less stiff airframe. Small HAVs can benefit from a lower proportion of airframe mass in terms of longer flight endurance and/or the ability to carry additional payload. It is an advantage if a small HAV can be provided with a less stiff airframe that is lighter in weight but does not lead to a reduction in performance due to increased vibration.

Market forces, including the need for portability, are a strong driver to the miniaturisation of HAVs and the lowering of their SWaP (size, weight and power), whilst maintaining most of the capability of larger HAVs.

The propulsion units in HAVs generate vibrations that excite the airframe in flight. A propulsion unit is typically a brushless, outrunner AC motor which directly drives a propeller. Typical excitation frequencies from the rotation speed of the propulsion units in a small HAV are 70-200 Hz (14 to 5 milliseconds (ms) per cycle), and harmonics of those frequencies. HAVs are typically used to carry cameras or other sensors for capturing images or other data, but the quality of the images or other data captured is adversely affected by vibration, so there is a need to compensate for vibrations caused by the propulsion units. Imagery for visual reconnaissance, such as video or stilt images, captured in-flight is higher quality if the camera on the HAV is not subject to vibration amplitudes that are high enough to move the camera axis significantly during a frame capture. Most high-resolution cameras have rolling shutters operating typically at 60 frames per second (16.7 ms rolling shutter time per frame) or less. High amplitude vibration of a rolling shutter camera results in a visually undesirable wobble distortion (a 'jelly' effect) on video imagery. This effect may be reduced with image post-processing before viewing, but in general, image processing introduces undesirable side-effects such as jagged lines and it is preferable to have the highest possible quality raw imagery so as to minimise or even do away with image processing.

Cameras for visual odometry are usually lower-resolution than reconnaissance cameras for visual imagery and usually have a global shutter, obviating the rolling-shutter 'jelly' effect. However, in low ambient light levels, the image exposure time can be 40 ms; in this case, high amplitude vibration of the camera results in pixel blur. Pixel blur reduces the effectiveness of the visual odometry by blurring the features, leading to a reduction in dead reckoning accuracy. Dead reckoning error is reduced in visual odometry with stereo cameras as opposed to a mono camera. The integration cost of stereo cameras includes the need to have a rigid stereo camera baseline and the need to reduce the vibrations experienced by both cameras. Conventionally this is achieved by mounting the stereo cameras on a rigid frame and soft mounting the rigid frame to the airframe to absorb vibrations. A common method of reducing the amplitude of vibration in a camera/gimbai is for the camera/gimbai to have significant mass and be mounted on soft vibration-absorption mounts ("soft dampers") from the main airframe and the driving oscillators of the propulsion units on the main airframe. In larger HAVs, weighing in the region of 750g, the main airframe with the propulsion units and battery comprise most of the weight. The camera of such a HAV, mounted on a 3-axis gimbal with 3 motors, still has significant weight and the camera-gimbal is mounted on soft dampers to absorb the vibration. This works well and the imagery captured in flight is of high quality. Intermediate HAVs weighing in the region of 300g may have a 2-axis gimbal with 2 motors and the weight of the soft dampers and support mounts/brackets for these is considerable. The most common type of soft dampers for HAVs are vibration damping balls which have a spherical shape between two flanges.

In contrast, in small HAVs, the gimbal is typically 1-axis and the total weight of camera(s) and 1-axis motorised gimbal is typically below lOg. This low camera-gimbal weight makes it challenging to achieve an effective reduction in vibration amplitude using a low-weight, soft mount from the main airframe with the driven oscillators of the propulsion units on it. We produced a prototype small HAV (weighing about 140-175g) with a minimal weight propulsion frame (the "propulsion frame" is part of the HAV airframe, and generally comprises a small central structure attached to rotor arms which support the motors and propellers for propelling the HAV) separated by a compliant interface from a main body (the "pod") containing the rest of the small HAV mass including the efectronics, the battery, the IMU (inertial measurement unit) and the camera-gimbal. In this prototype, the IMU fulfilled the functions of both sensing for flight control in the low-level loop and sensing for dead reckoning. The prototype had a soft compliant interface resulting in a low level of camera-gimbal vibration energy which was desirable. However, the interim prototype was uncontrollable in all but the least demanding flight conditions. It became clear during our work on this prototype that none of the solutions known about in the public domain provide sufficiently low levels of vibration on a small HAV for high quality imagery capture and accurate dead reckoning. There is a need for a small HAV which both minimises the vibration which would adversely affect the quality of the image and other data collected by the camera(s) and/or other sensor(s) carried by the HAV, whilst having robust flight control (so that the HAV is relatively stable and easy to control in flight).

SUMMARY OF THE INVENTION

The present invention is predicated on the realisation that a compliant interface that is close to a propulsion frame carrying the propulsion units will effectively damp all the sensors in the main body or pod of a small HAV. It was then realised that in our prototype, the compliant interface, which was soft, was introducing a large mechanical lag time between when the propulsion frame moved and when the IMU in the pod sensed that movement. This large, lag time in the interim prototype was often fatal for robust flight control and led to loss of control and crashes.

The present invention therefore provides a compact hovering air vehicle (HAV) comprising a propulsion frame compliantly mounted to a pod via a compliant interface, wherein a flight control Inertial Measurement Unit (IMU) is provided for control of the HAV in flight, the flight control IMU being mounted to the propulsion frame.

With such an arrangement, the compliant connection between the propulsion frame and the pod is sufficiently soft and provided close to critical damping such that negligible driven vibration energy from the propulsion frame is transferred through to the pod, thus enabling pod-mounted camera(s) and sensor(s) to collect quality images and data, minimally distorted by vibration. The term "propulsion frame" is used herein to denote the HAV propulsion unit(s) or motors and the minimal mass structure which holds the propulsion unit(s) in the flight position, which together are distinct from the pod containing the rest of the HAV mass. The provision of the flight control IMU on the propulsion frame allows a small mechanical lag time between movement of the propulsion frame and when that movement was sensed by the flight control IMU, and enables robust and responsive computer control of the propulsion units in a variety of flight conditions including hovering, full air velocity, sharp turns, rapid ascents/descents, rapid acceleration/deceleration in three dimensions and fast recovery from the imbalances of a collision without crashing to the ground. The compliant interface also ensures that only an acceptably low level of vibration energy is transferred to the pod when there is imbalance in at least one propulsion unit in the propulsion frame. The propulsion frame is designed to be as light as possible, with typically most of the weight of the HAV being contained in the pod; the pod is sufficiently heavy to damp the vibrations from the propulsion frame such that the sensors on the pod output sensor data of high enough quality for the small HAV to perform its task well. The mass of the pod, the mass of the propulsion frame, the stiffness and the damping of the compliant interface between the propulsion frame and the pod are determined, as a person skilled in the art would understand, so that there is not a natural frequency of oscillation between the pod and the propulsion frame during flight. Another advantage of the present invention is that the propulsion frame can be less stiff and lighter in weight than conventional designs without excessive vibration energy being transferred through to the pod.

The compliant interface suitably comprises a plurality of soft dampers, having both elastic and vibration damping properties. The soft dampers may be vibration damping balls. Alternatively the soft dampers can be springs, preferably helical springs but any shape spring or indeed any similar damping or spring-damping device that can be pressed or pulled but returns to its former shape when released, which is used to absorb movement without transmitting vibrations between the propulsion frame and the pod. Such soft dampers provide flexible movement along three orthogonal axes, and the flexibility may be greater afong one axis than the others (e.g. along the cylindrical axis of a helical spring). One of the orthogonal axes preferably extends directly between the propulsion frame and the pod, and this may coincide with the axis of greater flexibility of the soft damper.

There may be at least one motion limiter, the or each motion limiter being adapted to allow free motion of the propulsion frame relative to the pod in three dimensions, but to limit the extent of such movement in one or both directions in one, two or three dimensions. The motion limiter may limit motion in different directions and/or dimensions by different amounts. The presence of such motion limiters helps protect the soft dampers against over extension, or from damage such as when the HAV lands. A motion limiter can be integrated with a soft damper, which minimises the volume occupied by the compliant interface.

The compliant interface may be formed integrally with one of the propulsion unit and the pod; this minimises the HAV component count.

The flight control IMU may be rigidly mounted to the propulsion frame, or it may be compliantly mounted to the propulsion frame; in the latter case, the compliance may be provided by additional soft dampers. A second IMU (the "dead reckoning IMU") may be carried in the pod, for navigating the HAV by dead reckoning. One or more cameras may be carried in the main body, for capturing high quality imagery and/or for navigation by dead reckoning. Where more than one camera is used for dead reckoning it is important that the axes of the different cameras remain fixed relative to each other, otherwise the accuracy of the dead reckoning is adversely affected. Accordingly the cameras should be mounted to a rigid frame, such as a PCB; PCBs are usually planar, and are subject to thermal expansion in use, so the mounting of the PCB within the pod is arranged so as to allow for thermal expansion of the PCB without any bending of the PCB, since such bending would cause the camera axes to splay in/out, badly affecting dead reckoning accuracy.

Preferably the pod is heavier than the propulsion frame, which helps isolate the camera(s) from vibrations from the propulsion units.

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described by way of example and with reference to the accompanying figures, in which;

Figure 1 is a schematic isometric view of a hovering air vehicle (HAV) with an airframe mounting arrangement in accordance with the invention;

Figure 2 is an enlarged side elevation view of the airframe mounting arrangement of the HAV of Figure 1;

Figure 3 is a schematic view of the airframe mounting arrangement of Figure 2, and

Figure 4 shows a number of diagrams illustrating the modes of vibration of the airframe mounting arrangement in accordance with the invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

Figure 1 shows a generalised quad-rotor HAV 200 with four propulsion units, or motors 204, which are shown in a deployed configuration relative to the pod or main body 210 of the HAV; each propulsion unit is adapted to drive a propeller 206, and these propellers provide the HAV 200 with lift and propulsive force. The pod 210 is situated centrally between and below the four motors, is roughly prismatic and is longest in the X dimension. The propulsion units 204 are mounted at the distal ends of arms 202, and the proximal ends of the arms 202 are mounted to central frame portion 208, which is in turn mounted to the top of the pod 210 by a compliant mounting; this compliant mounting includes soft dampers 212 (more clearly shown in Figure 2) and motion limiters 214, and is described further below with reference to Figure 3. The central frame portion 208, the arms 202, the motors 204 and propellers 206 collectively form the HAV "propulsion frame".

Figure 3 shows the compliant mounting between the central frame portion 208 (the remainder of the propulsion frame is not shown) and the pod 210 of Figures 1 and 2 schematically in more detail. Four soft dampers in the form of helical springs 212 directly connect the central frame portion 208 and the pod 210, and there are four motion limiters 214 limiting the relative motion between the central frame portion 208 and the pod 210. Each motion limiter comprises a stop 216 enclosed in a limiting retainer 218, and the overall arrangement is such that the freedom of relative movement in six degrees of freedom (6-DOF) between the pod 210 and the central frame portion 208 (and hence the propulsion frame) is limited in extent - the motion limiters allow free movement of the propulsion frame relative to the pod between predetermined limits along an axis corresponding to the axis of the soft dampers, they also allow a limited amount of movement transverse to that axis (but not so much that the tips of one rotating propeller can come into contact with the tip of another propeller). Since the soft dampers 212 are soft, they can be damaged by over-stretching such as might occur in a hard landing or in rough handling. The stops 216 and limiting retainers 218 are provided to prevent damage to the soft dampers 212. The soft dampers 212 are designed with sufficient resistance to bending/movement in the plane transverse to the longitudinal axis of the helix so that the stops 216 and limiting retainers 218 are not in contact with each other when the HAV 200 is hovering or during normal flight manoeuvres. An optimum movement limit between the stops 216 and the retainers 218 for soft dampers 212 such that the stops 216 do not contact the retainers 218 as the soft dampers 212 flexed in most flight conditions is +/-lmm in all directions, but this limit could be greater or less than +/-lmm in some or all directions for HAVs of different size, weight, configuration or power. The aim of the compliant mounting formed by soft dampers 212 is to damp the vibrations from the propulsion frame such that the sensors on the (relatively much heavier) pod can output sensor data of high enough quality for the small HAV to perform its sensing task(s) well. 3D printing is used to reduce the component count, and hence the manufacturing cost, by 3D printing the four soft dampers 212, the stops 216 and retainers 218, where possible, on and integral with either the lower part of the central unit 208 or the upper part of the pod 210.

Figure 3 also shows a flight control IMU 220, in the form of a PCB with an IMU sensor mounted on it. The IMU sensor is preferably a MEMS type. The flight control IMU 220 is rigidly mounted in a carrier 222, and the carrier 222 in turn is attached to the central unit 208 by four carrier springs 224 forming another compliant interface. 3D printing is used to reduce the component count and hence the manufacturing cost by 3D printing the carrier 222, the four carrier springs 224 and the central frame portion 208 as one component. 3D printed springs may not always provide a compliant interface with sufficient vibration absorption properties and other compliant interface constructions can be used, as is well known to those skilled in the art. Wires 226 run from the flight control IMU 220 to the pod 210. The aim of this compliant interface is to permit signal information on the movement of the central unit 208 of the propulsion frame to be passed to the flight control IMU 220 on the carrier 222 without significant mechanical lag whilst removing extraneous vibration noise from the excitation of the propulsion frame that is detrimental to the quality of the signal. The compliant interface has stiffness and damping values that act as a high frequency filter (otherwise known as a low-pass filter) for damping vibrations of high frequency from the centra! unit 208 to the carrier 222. This results in lower levels of vibration noise from being sensed by the flight control IMU 220 without introducing significant mechanical lag time. Removing the high frequency noise sensed by the flight control IMU 220 improves the performance of the flight control low-level loop; however, the flight control IMU 220 could be rigidly mounted (directly) to the central unit 208.

The wires 226 that run from the flight control IMU PCB 220 to the pod 210 can affect the damping of the flight control IMU 220 on the carrier 222 relative to the propulsion frame. In a preferred arrangement, uitra-flexible, multi-strand copper wire coated in PTFE is used to minimise any damping effect of the wires 226 on the flight control IMU 220. There can be significant relative movement between the central unit 208 and the pod 210 -movement of the order of +/-lmm in 6-DOF. The route of the wires 226 is arranged to avoid a directional bias by splitting the wires into two bundles with symmetric routes. The importance and methods of designing and setting up a near-optimal provision of the stiffness and damping of the carrier springs 224 and wires 226 will be understood by a person skilled in the art. In another arrangement, the wires that run from the flight control IMU PCB 220 to the pod 210 are first damped to the central unit 208 of the propulsion frame before carrying on to the pod 210; this clamping removes any bias on the output of the flight control IMU 220 from movement between the flight control IMU PCB 220 and the pod 210.

The pod 210 has dead reckoning visual and inertial sensing integrated into it on a rigid frame 230. The rigid frame 230 has two cameras 232 with nominally parallel camera axes mounted to it. The baseline distance between the two 'stereo' cameras 232 is around 80-120mm but could be more than 120mm or less than 80mm. The rigid frame 230 has a dead reckoning IMU 228 rigidly mounted to it. This dead reckoning IMU 228 is a MEMS type IMU. The dead reckoning IMU 228 is primarily used as the inertial sensing component of a dead reckoning navigation system for the HAV 200.

The rigid frame 230 is preferably a PCB. The PCB is mounted in the pod 210 such that the PCB can expand/contract in a plane with changes in temperature without being subjected to bending stress. The preferred method of mounting is to support the PCB in a planar slot in a rigid pod 210 with room for planar thermal expansion. The PCB preferably has the dead reckoning IMU 228 and the two cameras 232 soldered onto it. The two cameras 232 are identical, have global shutters and are pointing downwards in normal flight. In this way, a visual-inertial dead reckoning system is provided that has: stereo vision, a rigid visual-inertial baseline, a sizable stereo baseline, global shutters, thermal robustness and a low vibration environment. With the advantages of this dead reckoning system, a dead reckoning error of less than 0.1% of distance travelled is achievable in a suitable visual environment. The PCB is preferably the Main PCB for the HAV 200 and dead reckoning is only one of its many functions.

The pod 210 has a gimbal unit 240 mounted to it which allows a camera 242 to rotate about a pitch axis 244 (shown as a point in Figure 3). If weight constraints permit, the gimbal unit could also allow the camera 242 to rotate about a roll axis (not shown in Figure 3, but shown figuratively in Figure 4) and/or about a yaw axis (not shown; all three axes are orthogonal). The pitch axis is adjacent to the camera 242, and the roll axis would be adjacent to the pod 210. The gimbal unit 240 is not vibration isolated from the pod 210 with soft mounts but instead is rigidly mounted to or integral with the pod 210. The pod 210 is vibration damped from the propulsion frame by one set of soft dampers 212. It is therefore not necessary to also compliantly mount the gimbal unit 240 and the dead reckoning rigid frame 230 to the pod 210. The result is a more compact and lighter weight HAV 200.

The pod 210 further comprises an energy store 250 such as one or more rechargeable lithium-ion battery cells. The energy store 250 is rigidly attached to the pod 210. The energy store 250 can be permanent and is recharged in situ or can be removed for recharging or swapped with a fully-charged replacement. There could be multiple energy stores 250 in the HAV 200 with different functions such as providing back-up during energy store swapping, or as a range extender. One or more energy stores could also be mounted on the propulsion frame. Preferably, there is a main energy store 250 that contains all or nearly all of the energy in the HAV 200, that is around 20-40% of the total weight of the HAV 200 at take-off and is mounted to the pod 210 increasing the mass of the pod 210, adding inertial damping to reduce the amplitude of vibration movement of the sensors (including the cameras) and increasing the quality of the data captured by the sensors.

Figure 4 is a diagram of the six simple modes of vibration isolated by the four soft dampers 212 between the propulsion frame and the pod 210. The four soft damper mounts on the propulsion frame are rigidly spaced with negligible relative movement between them and the four soft damper mounts on the pod 210 are rigidly spaced with negligible relative movement between them. The four soft damper mount centres are laid out on a rectangular footprint L x T and the soft dampers 212 have an active spring vertical height V. The six simple modes are three orthogonal rotational modes: Pitch, Roll and Yaw, and three orthogonal linear modes: longitudinal, transverse and vertical. Each linear mode has an associated stiffness which is a combination of the four soft dampers 212 acting together: Longitudinal (4*k|_), Transverse (4*kT) and Vertical (4*kv) where kL, kT and kv are the corresponding stiffnesses for one soft damper and the four soft dampers are identical. Each rotational mode has an associated torsional stiffness: Pitch (2*kv*L), Roll (2*kv*T) and Yaw is a function of L, T, k|_, kT.

The propulsion frame acts as a complex, driven oscillator to the pod 210 that is optimally isolated by the four soft dampers 212. The four propulsion units 204 at any one time during flight usually have different rotational speeds as the flight control low-level loop corrects for: the inherent under-actuation of the quad rotor in controlling 6-DOF with four actuators, the thrust/direction required for following the demand flightpath and the effect of any wind. The four changing speeds with four imbalances of the four propulsion units 204 generate multiple modes of vibration in the propulsion frame with different modes having different energies at different times.

The four soft dampers 212 are made of nylon which has both an elastic stiffness and a damping coefficient. Helical springs can be used where ki_= kT. It will be understood by those skilled in the art that the spring is specified (number of coils, diameter, filament section, material) to minimise the amount of vibration that transfers through from the propulsion frame to the pod 210.

In practice, an empirical approach to determining the best spring stiffnesses was adopted, several sets of soft dampers 212 with different stiffnesses were 3D printed and each set was tested in flight. During a test, the flight control IMU 220 measured the driving vibration energy in the propulsion frame and the dead reckoning IMU 228 in the pod 210 measured the received vibration energy in the pod 210. In testing, for an HAV of approximately 150g in weight, soft dampers 212 were found to be springs 3D printed from nylon with a 6mm diameter spiral (centre filament to centre filament), a 1.7mm diameter filament and a length of 2.5 turns worked well.

Further insight into reducing the vibration energy in the pod 210 was obtained with Fourier analysis of the recorded data from both IMUs 220, 228. Analysis of this data gave rise to the further possibility of using springs with asymmetric geometry such that kL is not equal to kT. The asymmetric spring may be ova! in XY geometric section instead of circular. In another arrangement, the asymmetric spring has a filament that varies in aspect ratio of filament width to filament height around a coil. The soft dampers 212 occupy volume in the design of the HAV and it is beneficial to minimise that volume. Helical springs waste volume in a tight, prismatic design. Springs that are rectangular in geometric section could be used instead, as these would be more compact than the equivalent helical springs.

In the drawings, the four stops 216 and four retainers 218 aiso occupy a different space to that occupied by the soft dampers 212, which adds to the overall volume of the HAV. The soft dampers 212, stops 216 and retainers 218 may be integrated to have a common axis, such as a vertical axis, thus reducing eight volumes to four volumes and increasing compactness.

The drawings show a HAV quad rotor with 4 arms. This invention is not limited to a quad rotor but includes HAVs with other numbers of arms - for example, a hex rotor with 6 arms or an octo rotor with 8 arms.

It will of course be understood that many variations may be made to the above-described embodiment without departing from the scope of the present invention. For example, this invention is not limited to simple propulsion arrangements. Instead, a pair of motors could be provided at the end of each arm with one propeller above the motors and the other propeller below the motors. Additionally or alternatively: the propulsion frame can be located below the pod instead of above the pod; the arms 202 can be slidable in the central unit 208 between deployed positions (as shown in Figure 1) and stowed positions (not shown); the arms 202 can be fixed to the central unit 208 or they can pivot from the central unit; the invention is particularly applicable to small HAVs weighing between 100-200g, but the applicability of the present invention includes HAVs weighing less than lOOg and more than 200g, and even heavier than 250g. The propellers may not ail be in the same approximate plane, so that at least one propeller swept area overlaps the swept area of at least one other propeller. One or more of the propeller axes could be tilted relative to other propeller axes such that each propeller axis is not parallel to every other propeller axis. The arms could be of different lengths, and/or the motors when deployed could be in a non-square configuration, with different sized gaps between the paths of the tips of adjacent rotating propellers. The propellers could be mounted so as to teeter, and/or the motors could be mounted to the distal ends of the arms on gimbals or the like so as to allow the orientation of the motors relative to the arms to be varied (or the propellers to teeter). Neither the motors nor the propellers need be of identical specification. Although an embodiment of the invention has been described which has an equal number (four) of propulsion units, soft dampers and motion limiters, there could be more or fewer than four soft dampers in the compliant interfaces and there could be more or fewer motion limiters than there are soft dampers. A soft damper in the compliant interface could comprise a series of spring-dampers between the propulsion frame and the pod, each spring-damper could have different stiffnesses and or damping coefficients. The dead reckoning system could have just one camera. The visual component(s) of the dead reckoning system could be of any number, or be of any combination of mono and stereo, each visual component could be at any angle to the HAV, and visual component(s) could be mounted on the gimbal. There can be multiple stereo dead reckoning fields of view, for example two cameras pointing vertically downwards and two cameras pointing horizontally. Any camera could capture imagery for multiple purposes including but not limited to: human viewing, dead reckoning, scene understanding and collision avoidance with static or moving objects. The frame PCB could be mounted other than by a slot such as on soft pillars such that the frame PCB does not bend with a change in temperature. The frame of the dead reckoning system could be something other than the PCB: for example, the frame could be the pod. The cameras and dead reckoning IMU could be attached to the pod separately. The mounting between the dead reckoning IMU and the frame could be compliant instead of rigid. The cameras could be camera modules with flexible connectors to connect them to other PCBs. The gimbal could carry more than one camera; each camera could be sensitive to the same wavelengths or different parts of the spectrum; a camera could be LWIR thermal or SWIR thermal; the cameras may be synchronised; image processing and compression may take place in the PCB in the gimbal; a third IMU can be provided on the camera PCB for gimbai control and or post-processing the imagery. Where different variations or alternative arrangements are described above, it should be understood that embodiments of the invention may incorporate such variations and/or alternatives in any suitable combination.

Claims (22)

1. A hovering air vehicle (HAV) comprising a propulsion frame compliantly mounted to a main body, or pod, via a compliant interface, wherein an Inertial Measurement Unit (IMU) is provided for control of the HAV in flight, the IMU being mounted to the propulsion frame.

2. A HAV according to Claim 1 in which the compliant interface comprises a plurality of soft dampers.

3. A HAV according to Claim 2 in which the compliant interface comprises at least one motion limiter, the or each motion limiter being adapted to allow free motion of the propulsion frame relative to the main body along orthogonal axes between predetermined limits.

4. A HAV according to Claim 3 in which the predetermined limits along one or more of the axes differ from those in the other axis/axes.

5. A HAV according to Claim 4 in which one of the axes is coincident with an axis of the soft damper.

6. A HAV according to any of Claims 3 to 5 in which the or each motion limiter is integrally formed with a soft damper.

7. A HAV according to any of Claims 2 to 6 wherein the or each soft damper is a spring in the form of a helical coil.

8. A HAV according to any preceding claim in which the compliant interface is formed integrally with one of the propulsion unit and the main body.

9. A HAV according to any of Claims 1 to 9 in which the IMU is compliantiy mounted to the propulsion frame.

10. A HAV according to any preceding claim in which a second IMU is carried in the main body.

11. A HAV according to Claim 10 in which the second IMU is rigidly mounted to the rigid frame.

12. A HAV according to Claim 10 in which the second IMU is compliantly mounted to the rigid frame.

13. A HAV according to any preceding claim further comprising one or more cameras carried in the main body.

14. A HAV according to Claim 13 in which the one or more cameras are rigidly attached to a rigid frame.

15. A HAV according to Claim 14 in which the rigid frame is a printed circuit board (PCB).

16. A HAV according to Claim 15 in which the PCB is planar and is mounted to the main body in a manner allowing limited movement of the PCB in its plane relative to the pod, so that the PCB does not bend with a change in temperature.

17. A HAV according to any preceding claim further comprising a gimbal mounted to the main body for allowing rotation of a sensing device about at least one axis.

18. A HAV according to Claim 17 in which the gimbai alfows rotation of a sensing device about a plurality of orthogonal axes.

19. A HAV according to Claim 17 or Claim 18 in which the mounting of the gimbal to the main body is rigid.

20. A HAV according to any preceding claim in which the main body is heavier than the propulsion frame.

21. A HAV according to any preceding claim in which the main body includes a main energy store mounted to the main body.

22. A HAV according to any preceding claim in which the main body includes most of the electronics and sensors mounted to the main body.