Payload carrying system and method

Abstract

A payload system for carrying a payload including a platform configured to support the payload, and a payload distribution sensor configured to sense the distribution of the payload on the platform. At least two autonomous aerial vehicles are provided, with coupling arrangements being configured to couple each vehicle to the platform. One or more processing devices configured to determine a payload distribution in accordance with signals from the payload distribution sensor and control operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted.

Description

PAYLOAD CARRYING SYSTEM AND METHOD Background of the Invention

[0001] The present invention relates to a system and method for carrying a load, and in one example, to a system and method for carry a load using multiple unmanned aerial vehicles (UAVs). Description of the Prior Art

[0002] The reference in this specification to any prior publication (or information derived from it), or to any matter which is known, is not, and should not be taken as an acknowledgement or admission or any form of suggestion that the prior publication (or information derived from it) or known matter forms part of the common general knowledge in the field of endeavour to which this specification relates.

[0003] Multirotor UAVs, also referred to as drones, known for their agility and compact size, are increasingly utilized as a prominent platform for air transportation and package delivery. Drones are typically designed for different payload capacities, and the total thrust limitations of individual drones restrict the payload they can lift. Using larger drones to carry high-capacity payloads can be unsafe for people and become too expensive and bulky, thus reducing their stability.

[0004] Cooperative aerial manipulation involves multiple drones working together to manipulate objects in the aerial environment, enabling tasks such as object transportation, assembly, and manipulation with enhanced capabilities and coordination. Cooperative UAVs have emerged as a viable solution for handling heavy or oversized payloads that surpass the capabilities of individual robots, as described in Lee, H. and Kim, U. 2021. Estimation and Control of Cooperative Aerial Manipulators for a Payload with an Arbitrary Center-of-Mass. Sensors 2021, Vol.21, Page 6452.21, 19 (Sep.2021), 6452.

[0005] Experimental systems for cooperative transport using drones, typically employing two to four drones, have commonly implemented a lift or pull-based method, positioning the drones above or beside the load being transported. These pull-based paradigms all exhibit similar disadvantages, namely, the payload underneath the drones is subject to strong downwash fromthe drone propellors, typically inducing a pendulum swing that can further disbalance the payload, causing additional strain on the cooperative drones, potentially damaging the payload, and inducing intense vibrations on the drone. At the same time, wind may perturb the balance and introduce load oscillation as well. Additionally, scenarios involving transporting comparatively larger objects might present slipstream issues due to the interference with the propellers’ downwash airflow.

[0006] While cooperative, mobile manipulators have been developed to exploit grasping capabilities, their practical implementation is hindered by the complexities associated with multiple aerial robots as discussed for example in Kellermann, R. et al.2020. Drones for parcel and passenger transportation: A literature review. Transportation Research Interdisciplinary Perspectives. 4, (Mar. 2020). Previous approaches have primarily focused on addressing control and coordination challenges by assuming a uniform mass distribution in the payload, simplifying the problem. However, manipulating payloads with non-uniform mass distributions, where the payload's geometry and center of mass differ, introduces significant complexities.

[0007] In contrast to the scenario of a single robot, effectively coordinating a team of multiple mobile platforms in sync necessitates additional factors to be considered. These include ensuring synchronized motion among the agents, effectively managing the stresses exerted on the manipulated object, and maintaining the overall stability of the combined system. This is discussed in Bacelar, T. et al.2020. On-board implementation and experimental validation of collaborative transportation of loads with multiple UAVs. Aerospace Science and Technology. 107, (Dec. 2020), 106284 and Mellinger, D. et al. 2011. Design, modeling, estimation and control for aerial grasping and manipulation. (Dec.2011), 2668–2673.

[0008] As a result, cooperative manipulation has garnered significant attention in research, primarily focusing on stationary or ground-based mobile manipulators.

[0009] In multi-vehicle lift and payload transportation, a sling load configuration offers advantages over a rigidly attached configuration where vehicles are directly connected to the payload. The sling load configuration allows for greater separation between the vehicles and the payload, introducing rotational degrees of freedom that reduce dynamic constraints in the system. However, while using tethers reduces constraints on the vehicles and payload, itintroduces multiple vibrational degrees of freedom that can be excited during maneuvers, making control design more complex, especially when the payload's mass, inertia, and aerodynamic properties are uncertain. Additionally, tethers bring the risk of in-flight collisions between vehicles.

[0010] Another approach in prior work involves multiple UAVs rigidly attached to a payload. However, these methods often rely on strict constraints on vehicle geometry or strong assumptions about precise knowledge of payload properties, see for example Raptopoulos, A. et al. 2013. Transportation using network of unmanned aerial vehicles and Rinaldi, J. et al. 2017. Package delivery mechanism in an unmanned aerial vehicle.

[0011] In Wang, Z. and Schwager, M. 2016. Kinematic multi-robot manipulation with no communication using force feedback. Proceedings - IEEE International Conference on Robotics and Automation.2016-June, (Jun.2016), 427–432 and Wang, Z. and Schwager, M. 2015. Multi-robot manipulation with no communication using only local measurements.2015 54th IEEE Conference on Decision and Control (CDC). IEEE.380–385 predefined multi-agent geometries are considered, which have limited applicability in cases with available or unknown attachment geometries.

[0012] Franchi, A. et al.2016. Distributed Estimation of State and Parameters in Multi-Agent Cooperative Load Manipulation. IEEE Transactions on Control of Network Systems.6, 2 (Feb. 2016), 690–701, proposes a two-phase estimation scheme for multi-agent cooperative lift systems with uncertain parameters, dividing the process into kinematic and dynamic phases. While this work represents a significant advancement in adaptive control for these types of systems, it has limitations as it does not address estimation during takeoff (i.e., during ground contact) and requires each agent to estimate its local velocity in the parent frame, which may be challenging to achieve in practical scenarios.

[0013] In aerial manipulation and cooperative tasks, various attachment methods have been studied. These include two primary categories of multi-agent aerial manipulators and their combinations, which are distinguished by how the UAVs are interconnected to form a floating structure. Current technologies have heavily emphasized the use of wires in this circumstance. However, methods and controls provide alternatives, such as hanging cables or rigid connections made by various flexible linkages. Webb, K. and Rogers, J. 2021 "Adaptivecontrol design for multi-uav cooperative lift systems", Journal of Aircraft. 58, 6 (Jul. 2021), 1302–1322 present a novel adaptive control design for multiple UAVs to lift objects with a 3, 4, or 5-quadcopter approach. This shows that using a four-drone approach with rigid attachments for adaptive control is a viable solution to the swarm lift approach. However, existing solutions typically employ a pull approach to lifting.

[0014] Designing an additional object on an existing quadrotor presents challenges, such as ensuring proper weight distribution, structural integrity, and compatibility with the drone's control systems and aerodynamics. Robotic arm attachment has been widely studied in the literature due to its potential to enhance the versatility and functionality of quadrotors, enabling them to perform complex manipulation tasks and interact with their environment more effectively. These approaches can potentially be used in inspecting, manipulating, or transporting objects in remote places, and whilst simulation findings are encouraging, their practical use requires multi-axis force / torque sensors to overcome coordination concerns and successfully manage unknown payloads. In other approaches, a rapid aerial grasping method to pick up and transport objects with a soft robotic gripper has been used. However, the developed system consists of the robotic 9DoF arm that uses IMU sensor readings to accurately maintain the object in position and manage the added load on the drone.

[0015] US9043052 provides a system and method for controlling a plurality of vehicles to affect positioning of a common payload. The system comprises of multiple vehicles having positioners to change the location of the common payload, where the group of vehicles form a swarm that is controlled by a driver or pilot station. Each vehicle is autonomously stabilized and guided through a swarm electronics unit, which further includes sensor, communication, and processing hardware. At the driver or pilot station, a system or a person remotely enters payload destinations, which is processed and communicated to each vehicle. The method for controlling a multi-vehicle system includes inputting the desired location of the payload and determining a series of intermediary payload waypoints. Next, these payload waypoints are used by the swarm waypoint controller to generate individual waypoints for each vehicle. A controller for each vehicle moves the vehicle to these individual waypoints.

[0016] US20150251756 describes a system and method for commanding a payload of an aircraft. A plurality of flight segments, which comprise trajectory information of the aircraft,are received. A plurality of payload commands are generated using statements of payload intents. Each one of the payload commands are synchronized with at least one of the plurality of flight segments. The system and method express the operations to be performed by the payload onboard in order to achieve the established mission goals of the aircraft.

[0017] US20180188724 describes systems and methods using a drone swarm to increase cargo capacity. A drone swarm may include a networked drone system or two or more drones, such as a parent drone and a child drone. A method may include receiving support component balance information captured by an inertial measurement unit on the support component supported by a parent drone, adjusting movement of the parent drone according to a control system using the support component balance information, receiving an indication of a low battery in a drone in the networked drone system, the indication including an identification of a replacement drone to replace the drone with the low battery in the networked drone system, and sending a reconfiguration command to at least one child drone to incorporate the replacement drone in the networked drone system.

[0018] However, none of the above solutions adequately solve the issue of load transport using multiple drones, particularly when transporting uneven and / or loads with a moving centre of mass, such as loads containing fluids. Summary of the Present Invention

[0019] In one broad form, an aspect of the present invention seeks to provide a payload system for carrying a payload, the system including: a platform configured to support the payload; a payload distribution sensor configured to sense the distribution of the payload on the platform; at least two autonomous aerial vehicles; at least two coupling arrangements, each coupling arrangement being configured to couple a respective one of the vehicles to the platform; and, one or more processing devices configured to: determine a payload distribution in accordance with signals from the payload distribution sensor; and, control operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted.

[0020] In one broad form, an aspect of the present invention seeks to provide a method for carrying a payload, the method including: providing: a platform configured to support thepayload; a payload distribution sensor configured to sense the distribution of the payload on the platform; at least two autonomous aerial vehicles; at least two coupling arrangements, each coupling arrangement being configured to couple a respective one of the vehicles to the platform; and, in one or more processing devices: determining a payload distribution in accordance with signals from the payload distribution sensor; and, controlling operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted.

[0021] In one embodiment the coupling arrangements are configured to allow an orientation of each vehicle relative to the platform to be adjusted.

[0022] In one embodiment the orientation includes the pitch, yaw and roll of each vehicle relative to the platform.

[0023] In one embodiment each coupling arrangement includes actuators for controlling an orientation of the respective vehicle relative to the platform and wherein the one or more processing devices are configured to adjust the orientation of each vehicle in accordance with the determined payload distribution.

[0024] In one embodiment each coupling arrangement includes three actuators for controlling a pitch, yaw and roll of the respective vehicle.

[0025] In one embodiment the actuators include rotational servo motors.

[0026] In one embodiment each vehicle includes an altitude sensor, and wherein the one or more processing devices are configured to: determine vehicle altitudes in accordance with signals from the altitude sensors; and, control operation of the vehicles in accordance with the determined altitudes.

[0027] In one embodiment each altitude sensor includes at least one of: a downward facing ranging device; and, a downward facing LiDAR.

[0028] In one embodiment the payload distribution sensor includes a resistive panel.

[0029] In one embodiment the payload distribution sensor includes orthogonally arranged sensing elements.

[0030] In one embodiment each coupling arrangement is magnetically attached to a respective vehicle.

[0031] In one embodiment the platform is situated above a centre of lift of each vehicle.

[0032] In one embodiment the one or more processing devices are configured to control the vehicles while carrying the load so as to at least one of: control an orientation of the platform; maintain the platform in a balanced position; maintain the platform substantially level while carrying the payload; and, maintain forces on the load substantially normal to a plane of the platform.

[0033] In one embodiment the one or more processing devices are configured to: determine a destination location; and, control the vehicles to transport the platform and payload to the destination location.

[0034] In one embodiment the one or more processing devices interact with a flight controller of at least one of the vehicles.

[0035] In one embodiment the one of the vehicles is a leader vehicle and other ones of the at least two vehicles are follower vehicles.

[0036] In one embodiment the system is configured to implement a path planning algorithm to determine a path to the destination.

[0037] In one embodiment the path planning algorithm incorporates asynchronous angle adjustments to resolve the effect of asynchronous angles between the leader and follower vehicles. In one broad form, an aspect of the present invention seeks to provide a method for carrying a payload, the method including: providing: a platform configured to support the payload; a payload distribution sensor configured to sense the distribution of the payload on the platform; at least two autonomous aerial vehicles; and, at least two coupling arrangements, each coupling arrangement being configured to couple a respective one of the vehicles to the platform; and, in one or more processing devices: determining a payload distribution in accordance with signals from the payload distribution sensor; and, controlling operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted.

[0038] It will be appreciated that the broad forms of the invention and their respective features can be used in conjunction and / or independently, and reference to separate broad forms is not intended to be limiting. Furthermore, it will be appreciated that features of the method can be performed using the system or apparatus and that features of the system or apparatus can be implemented using the method. Brief Description of the Drawings

[0039] Various examples and embodiments of the present invention will now be described with reference to the accompanying drawings, in which: -

[0040] Figure 1A is a schematic side view of a payload system for carrying a payload;

[0041] Figure 1B is a schematic plan view of the payload system of Figure 1A;

[0042] Figure 2A is a schematic diagram of an example of a processing system for use in the payload system of Figure 1A;

[0043] Figure 2B is a schematic diagram of an example of a flight controller for a vehicle used in the payload system of Figure 1A;

[0044] Figure 3 is a flow chart of a process for carrying a payload;

[0045] Figure 4 is an image of an example of a payload system;

[0046] Figure 5 is a schematic diagram of an example of a control system;

[0047] Figure 6A is an image of an example of a vehicle attached to an actuated coupling;

[0048] Figure 6B is close up image of the actuated coupling of Figure 6A;

[0049] Figure 6C is an example of the configuration of the degrees of freedom for the actuated coupling of Figure 6A;

[0050] Figure 7 is an image of an example of a payload system showing a coordinate frame for path planning;

[0051] Figure 8A is a schematic diagram of an example of a grid arrangement for a payload distribution sensor;

[0052] Figure 8B is a heat map of an example of servo pitch angle for different payload distribution;

[0053] Figure 8C is a graph of examples of controller performance for yaw movements;

[0054] Figure 8D is a graph of examples of controller performance for pitch movements;

[0055] Figure 8E is a graph of examples of controller performance for roll movements;

[0056] Figure 9A is a graph illustrating examples of flight vibration results for a system with a 3 degree of freedom actuated coupling;

[0057] Figure 9B is a graph illustrating examples of flight vibration results for a system without a 3 degree of freedom actuated coupling;

[0058] Figure 10A is an image of an example experimental configuration for 3-axis angle measurement;

[0059] Figure 10B is an image of an example experimental configuration for vehicle angle measurement;

[0060] Figure 11A is a schematic diagram of a first side view of platform coordinate and moving frames;

[0061] Figure 11B is a schematic diagram of a second side view of platform coordinate and moving frames;

[0062] Figures 11C is a schematic diagram of a plan side view of platform coordinate and moving frames;

[0063] Figure 12A is an image of an example of vehicle movement with a payload with rotational angles of ψ_A=ψ_B=45°;

[0064] Figure 12B is an image of an example of vehicle movement with a payload with rotational angles of ψ_A=ψ_B=15°;

[0065] Figure 13 is a graph of example results of platform angle relative to vehicle attitude;

[0066] Figures 14A and 14B are graphs of example platform motion results for asynchronous vehicle angles;

[0067] Figures 15 is a graph of example platform motion results for vehicle angle alteration;

[0068] Figure 16A is a schematic diagram of an example coordinate frame for the payload distribution sensor;

[0069] Figure 16B is a schematic diagram of an example of actuator angle deviation for no payload;

[0070] Figure 16C is a schematic diagram of an example of actuator angle deviation for a small payload;

[0071] Figure 16D is a schematic diagram of an example of actuator angle deviation for a large payload;

[0072] Figure 17 is a graph of example platform motion results for different vehicle attitudes; and,

[0073] Figure 18 is a graph of example platform motion results for vehicle attitude alteration. Detailed Description of the Preferred Embodiments

[0074] An example of a system for system for carrying a payload will now be described with reference to Figures 1A and 1B.

[0075] The system includes at least two UAVs or other similar vehicles 100A, 100B, with two being shown in this example of the purpose of illustration only. The exact form of the vehicles will vary depending on a range of factors, such as the nature of the payload and the distance over which the payload is being carried, but typically the vehicles are multi-rotor drones or similar.

[0076] Each vehicle 100A, 100B is attached via a coupling 111 to a platform 110, which is configured to support a payload L. The nature of the payload can vary depending on the intended usage, but notably is not limited to solid objects and could for example encompass loose objects, as well as containers containing a fluid, or a fluid directly, assuming the platform is suitably shaped.

[0077] The platform includes a payload distribution sensor 112 configured to sense the distribution of the payload on the platform. The nature of the sensor will vary depending on the preferred implementation, and could include multiple load cells, for example arranged in an array, or could include touch sensitive resistive panels, or similar. It will also be appreciated that a combination of sensors could be used, for example combining loads cells to sense an overall weight, with a panel or array of sensors being used to sense a weight distribution over the platform.

[0078] The system typically includes one or more processing devices, which may form part of one or more processing systems, or the like. In one example, this can incorporate flight control systems of the vehicles as well as one or more separate system controllers, as will be described in more detail below. Whilst the system can use multiple processing devices, with processing performed by any one or more of the devices, it will be appreciated that alternatively a single processing device or system could be used. For the purpose of ease of illustration the following examples will refer generally to processing devices, but it will be appreciated that reference to a singular processing device should be understood to encompass multiple processing devices and vice versa, with processing being distributed between the devices as appropriate.

[0079] In use, the one or more processing devices are configured determine a payload distribution in accordance with signals from the payload distribution sensor and then control operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted.

[0080] By using signals from the payload distribution sensor, this allows the system to adapt control of the vehicles to the particular configuration of the payload. For example, if the payload is offset from a centre of the platform, this can be accommodated by generating additional lift from one of the vehicles, thereby ensuring the overall system remains balanced.

[0081] Additionally, this can take into account changes in the weight distribution during flight. For example, as the vehicles and platform perform manoeuvres, the changing in forces can alter the weight distribution, for example increasing the downward force on one side of the platform if the system turns during flight. This can also accommodate changes in distribution, for example if the payload moves on the platform, or if there is a change in external forces, such as wind loading. Additionally, this can be used to accommodate changes whilst forces on the load can change the distribution even if the load remains static, for example if the payload includes a liquid in a container. In any of these cases, changes in load distribution will change the required flight characteristics of the vehicles, and so by controlling operation of the vehicles based on the weight distribution, this can allow such events to be accommodated.

[0082] A number of further features will now be described.

[0083] A specific example of the one or more processing devices as part of a control system will now be described with reference to Figures 2A and 2B.

[0084] In this example, the control system includes at least one system controller 220, which is configured to analyse signals from the payload distribution sensor 112, as well as vehicle flight controllers 230.

[0085] The system controller 220 typically includes at least one processing device, such as a microprocessor 221, connected to a memory 222, an input / output interface 223, and a sensor interface 224. The sensor interface 224 is used for connecting the processing device 221 to the payload distribution sensor 112, and optionally any other sensors, such as altitude sensors, or the like.

[0086] The input / output interface 223 can be used to receive commands from an external system, such as a radio control unit, computer system, or the like. This enables external commands to be provided to the system, for example allowing the system to be controlled to allow an operator to provide an indication that carrying of the payload can commence, provide an intended destination for the payload, or the like. The input / output interface 223 can also be used to interconnect multiple system controllers 220, to connect the system controller 220 to one or more of the flight controllers 230, and may also connect to one or more coupling actuators, as will be described in more detail below. Although a single interface is described,in practice functionality may be provided by a number of distinct input / output interfaces 223, such as a serial interface, a network interface, wireless interface, or the like.

[0087] In use, the processing device 221 executes applications software stored in the memory 222, allowing the processing device 221 to perform necessary functions, including receiving data from the payload distribution sensor 112, processing and interpreting the data, and generating instructions for the flight controllers 230.

[0088] The flight controllers 230 typically include at least one processing device, such as a microprocessor 231, connected to a memory 232, an input / output interface 233, a sensor interface 234, and a drive system 218.

[0089] The sensor interface 234 is used for connecting the processing device 231 to one or more onboard sensors. The sensors could include an altitude sensor 235, which in one example is a downwardly pointing ranging device, such as a LiDAR, a positional sensor 236 such as a Global Positioning System (GPS) Sensor, and a motion sensor 237, such as an Inertial Measurement Unit (IMU). However, it will be appreciated that a range of other sensors could be provided, such as a ranging device in the form of a LiDAR, Radar, stereoscopic camera or the like, and the example arrangement is for the purpose of illustration only. It will be noted that each imaging device 235 is typically movably mounted to the respective vehicle, for example using a gimbal mounting or similar, and that the processor 231 is also typically configured to control operation of the gimbal and hence the orientation of the imaging device relative to the vehicle.

[0090] The nature of the drive system 218 will depend on the particular vehicle, but as mentioned above, this typically includes a number of rotor motors, for example in the case of a multi-rotor drone. The nature of the drive system is not important for the purposes of the current explanation and it will be appreciated that a wide range of different arrangements could be employed depending on the preferred implementation.

[0091] The input / output interface 233 can be used to communicate with the system controller 220, and / or could be used to provide connectivity to other external controllers, computer systems, or the like. This enables commands to be provided to the vehicle flight controller 230 to allow the vehicle to be controlled. For example, this could include providing generalcommands, such as instructing the vehicle to fly along a particular heading, with the processing device 231 then generate the necessary control commands to control the drive system 238 as required.

[0092] In use, the processing device 231 executes applications software stored in the memory 232, allowing the processing device 231 to perform necessary functions, including receiving data from the sensors 235, 236, 237, processing and interpreting the data, and making any required decisions. The processing device can also receive input commands from the system controller 220 and generating control instructions for the drive system 218.

[0093] Additionally, processing could be distributed between the processing devices of different vehicles as needed. For example, system controllers 220 could be provided for each vehicle, so that each vehicle 100A, 100B includes a system controller 220A, 220B that controls the flight controller of that vehicle. Alternatively a single system controller 220 could be used to control multiple vehicles, for example, the vehicles could be configured in a master slave configuration with the system controller 220 controlling a first one the vehicles 100A by interfacing directly with the flight controller 230A of the first vehicle 100A, whilst the other vehicle 100B is slaved to the first vehicle 100A, so that the flight controller 230B of the second vehicle 100B receives commands from the first flight controller 230A. This can assist with scaling the system making it easier to employ larger numbers of vehicles.

[0094] In one example, the coupling arrangements 111 are configured to allow an orientation of each vehicle 100A, 100B relative to the platform to be adjusted. This could include for example using a gimbal mechanism, or other similar arrangement, to enable the vehicle orientation to be adjusted, which in one example includes altering the pitch, yaw and roll of each vehicle 100A, 100B relative to the platform. Being able to adjust the vehicle orientation in this manner can help ensure the system is able to lift the platform taking into account the current distribution of the load.

[0095] The coupling arrangements 111 could be passive, allowing the orientation of the vehicles to be adjusted using the vehicles own flight control systems, so that each vehicle could rotate to adjust a yaw relative to the platform. However, this places an additional burden on control of the vehicles, which can lead to system instability and vibration. Accordingly, more typically each coupling arrangement 111 includes actuators for controlling an orientation ofthe respective vehicle relative to the platform. In this case, the processing devices can be configured to adjust the orientation of each vehicle in accordance with the determined payload distribution, with this functionality typically being performed by having the system controller 220 control the actuators.

[0096] In this example, each coupling arrangement 111 typically includes three actuators for controlling a pitch, yaw and roll of the respective vehicle relative to the platform. Whilst the nature of the actuators could vary depending on the preferred implementation, in one example the actuators can include rotational servo motors, although linear actuators, or similar actuators could be used with suitable configuration.

[0097] In one example, each vehicle includes an altitude sensor 235, with the processing devices, being configured to determine vehicle altitudes in accordance with signals from the altitude sensors and control operation of the vehicles in accordance with the determined altitudes. The altitude sensors could be of any appropriate form, and could include downward facing lidar, or radar, or other suitable range sensors.

[0098] In one example, the payload distribution sensor includes a resistive panel, and in one particular example, a resistive panel with orthogonally arranged sensing elements, allowing the load distribution to be determined, and further details of an example arrangement will be described in more detail below.

[0099] In one example, each coupling arrangement is magnetically attached to a respective vehicle. This allows the platform to be easily decoupled from the vehicles when not in use, in turn facilitating interchange of vehicles, although it will be appreciated that other coupling arrangements could be used.

[0100] In one example, the platform is situated above a centre of lift of each vehicle, with the coupling arrangement being attached to an upper surface of the vehicles, and more typically above the rotors of a multi-rotor vehicle. This arrangement allows a push configuration to be used to, so that the vehicles are pushing the platform upwards from below. This configuration has a number of advantages. For example, this allows the payload to be lifted and landed more easily, avoiding the issue of the payload contacting the ground before the vehicles, which can cause issues during take-off and landing with suspended payloads. This also allows the systemto ensure the payload is balanced and supported during the take-off process, rather than having the vehicles already in flight when this is attempted. Additionally, this can reduce interference between the load and platform and downwash from the rotors, which in turn can lead to flight instability, whilst also allowing for improved kinematics and weight distribution analysis, leading to improved control outcomes. Nevertheless, it will be appreciated that the techniques described herein could be configured for use with a pull configuration, in which the platform is attached beneath the vehicles.

[0101] In one example, the processing device is configured to control the vehicles while carrying a load so as to control an orientation of the platform. Typically this is performed to maintain the platform in a balanced and / or level position, so as to prevent the load moving on the platform, and particularly falling off the platform. However, it will be appreciated that while manoeuvring, this could include maintaining forces on the load substantially normal to a plane of the platform, for example by banking the platform while turning, for similar reasons.

[0102] As mentioned above, the processing devices are typically configured to determine a destination location and control the vehicles to transport the platform and payload to the destination location. Specifically, in one example, a system controller is used to interact with a flight controller of one or more of the vehicles, with the one of the vehicles typically acting as a leader vehicle and other ones of vehicles being follower vehicles, so that the system controller can interact with the flight controller of the leader vehicle.

[0103] The destination could be provided as a location, with the vehicles and platform operating autonomously to travel to the location. Alternatively, the destination could be determined based on user input commands, for example provided by a remote radio controller, or other similar device.

[0104] In one example, the system, and more typically the system controller, is configured to implement a path planning algorithm to determine a path to the destination, with the system controller being configured to instruct the flight controllers of the vehicles based on the path. The path planning algorithm typically incorporates asynchronous angle adjustments to resolve the effect of asynchronous angles between the leader and follower vehicles, which in turn can be used to determine the required orientation of the vehicles relative to the platform, using this information to control actuators in the case of active coupling arrangements.

[0105] An example of the control process for the system will now be described with reference to Figure 3.

[0106] In this example, at step 300, the system controller 220 determines a destination and uses this to calculate a path to the destination at step 310. This process involves using an understanding of the relative physical positioning of the vehicles 100A, 100B relative to the platform, and hence each other, in order to calculate the necessary angle adjustments to the yaw of the vehicles during flight.

[0107] At step 320, the system controller 220 determines a current load distribution from the load distribution sensor 112, and simultaneously determines vehicle altitudes at step 330, from the onboard altitude sensor 235. The system controller 220 then calculates required coupling configurations at step 340, before controlling the actuators and vehicles at step 350. Steps 320 to 350 are then repeated continuously as the vehicles transport the payload to the destination.

[0108] Details of specific arrangements will now be described in more detail.

[0109] In one example of the current approach a ‘push’ rather than a ‘pull’ approach is used. Specifically, in one example, the approach uses a rigid body attachment with servo motor control that uses the ‘push’ mechanism attached to the top of the vehicles, which in this example are multirotor drones, rather than the sides or bottom of the vehicle. This approach helps with analysis of the weight distribution and kinematics experienced by the drones while performing cooperative aerial manipulation tasks, and in turn allows for better control. SYSTEM MODELLING

[0110] Figure 4 illustrates a specific example of two drones lifting together, with similar features to those described above being denoted by similar reference numerals, albeit with the prefix 400. Thus, the system includes two drones 400A, 400B attached via couplings 411 to the platform 410. The system includes two system controllers 420, with a respective flight controller 430 being provided for each drone. Each drone includes a LiDAR 435 and position sensor 436.

[0111] In this example, the drones attach to the platform via magnetic couplings, which are flexibly designed to be easily detachable so that the arrangement can be readily adapted to moststandard multirotor drones and, in the future, can be readily extended to a higher number of drones cooperating in joint lift and transport. The platform 410 contains sensors that relay the balance at different points on the resistive panel in the tray to the drones. The couplings 411 each contain 3-degree end effector (3-DEE) actuators / motors that can adjust the tray to maintain the balance of the shared payload. The drones are arranged in a leader-follower configuration, where the lead drone receives the tray sensing data and instructs the follower drone how to actuate the tray to maintain balance while adjusting its actuators.

[0112] Figure 7 visually represents the relationship between total thrust, weight, and object load. The leader-follower drones use the 3 DoF servo to control movement of the platform 410, hereinafter referred to as a Self-Balancing Tray (SBT). The altitude for both drones is derived using downward-facing LiDARs 435. Figure 7 depicts the placement of the SBT 410 and the correlation between the total thrust generated by the two drones and the weight they carry. If the object's weight increases, lifting it with a single drone is impossible. The total thrust required to lift and sustain that load must also increase; using drones in a swarm will resolve this constraint.

[0113] The drones are spatially aware of the payload on the SBT, and adjust the thrust output in response to the changing weight to maintain stability and lift the load effectively. The table below references configurations of the drone developed for testing purposes. The payload can vary depending on the application or purpose of the drone system, ranging from small packages or equipment to larger items. The table shows both comparisons for the Thrust ability with 2- blade and 3-blade propellers. Both drones can lift a maximum of 2.5 kg. Table 1: The drone hardware specifications

[0114] This following section further discusses the unique SBT model design and the control architecture of the drones in a swarm. Drone Architecture

[0115] Figure 5 shows the swarm drones' configuration for the intended application. The NVIDIA Jetson Nano is the leader drone's system controller 520, hereinafter referred to as an onboard computer (OBC). The OBC 520 commands the flight controller 530, which controls the drone's flight modes and trajectory. By interacting with the flight controller 530 via MAVLINK, the OBC 520 secures the drone's flying mode selection and location. The flight controller connects the flight controls, such as ESC (Electronic Speed Control) and motors, to the output from the positioning sensors like GPS, IMU, and LiDAR. The payload distribution sensor 512 includes a resistive panel arrangement that communicates with a microcontroller 540 of the leader drone, assisting in localizing the payload on the SBT.

[0116] The microcontroller is an Arduino Nano 33 IoT, which also controls the actuators, which in this example include servos to provide 3 DoF for the SBT while communicating with the OBC via a two-way serial line. The Arduino board also controls the servo motors and enables wireless position data transfer between leader and follower drones via BLE (Bluetooth Low Energy), ensuring synchronized control of the SBT.

[0117] Figure 5 also aims to highlight the system procedures for the SBT architecture. Once the target position is acquired, the leader drone initiates the process by sending accurate position coordinates for the SBT with accurate GPS coordinates. The follower drone system accepts the signal request and transmits the verification signal to confirm communication. With confirmed verification, the drones in the swarm lift the payload and deliver it from point A to B. SBT Model

[0118] The Self-Balancing Tray (SBT) incorporates three servo motors S1, S2, and S3, with torque control to precisely control over the drone's yaw, roll, and pitch angles, respectively.The limitations of the angle configuration depend on the developed design and hardware constraints, as seen in Figures 6A to 6C. The reason for these limitations is typical to ensure safe and stable operation, preventing excessive angles that may compromise the drone's maneuverability or structural integrity.

[0119] In practice, the drones lock altitude using an onboard LiDAR ranging sensor, providing a coarse-grain control of the system, while the 3-DEE actuator system provides a fine-grain control of the payload. The 3-DEE system maintains the centre of mass of the payload’s varying positions.

[0120] The two drones' configurations are similar, with one serving as the leader (master) and the other as the follower (slave). The Arduino board controls the servo motors, providing a reliable and programmable control interface. The position data transfer between the leader and follower drones is achieved wirelessly, utilizing BLE (Bluetooth Low Energy) due to their short-range communication capabilities. The leader transmits the reference position of the servo to ensure similar control of the SBT in the follower drone. Both drones connect to SBT using magnets to facilitate easy attachment and detachment. The scalable design will accommodate more than two drones for more extensive operations in future implementations.

[0121] The design focuses on developing a non-linear control system for the three degrees of freedom (3 DoF) based on sensor fusion for the SBT. This sensor fusion integrates information from accelerometers, gyroscopes, and external sensors to accurately determine the drone's orientation and angular movements for the SBT. As the previous section shows, the SBT incorporates two external resistive panels, ensuring high-resolution shifting and positioning measurements of the payload.

[0122] The resistive panel principle has a gap separating top and bottom transparent conductive sheets with uniform resistance values. When there is contact on the top sheet, the touched point gives and contacts the bottom sheet. The location is determined based on the pressure on the contact point. One sheet includes electrodes for the vertical direction, and the second has electrodes for the horizontal direction, a feature of analog 4-wire resistive technology. The other sheet measures the voltage of a touched location. Therefore, the contacted location in the X and Y directions is determined using the analog outputs. Arduino's internal analog-to-digital converter (ADC) converts analog readings from the resistive panel into digital data.

[0123] The SBT employs a grid system to precisely monitor the tray's position and the target object's alignment. The resistive panel must account for the shifting caused by the drone's movements and the weight and friction between the target object and the tray. Fusion sensors monitor the drone's angle to enhance stability and alignment further and ensure that the tray remains level. Additionally, the resistive sensor ensures that the target object remains centered on the tray throughout the delivery process. CONTROL SYSTEM

[0124] The control system helps in successfully operating drone payload delivery systems. It enables precise maneuverability, stability, and control over drones during transportation. This section discusses designing and implementing a comprehensive control system for our proposed drone payload delivery solution. The control system incorporates various components, including the self-balancing tray (SBT), the three degrees of freedom (3 DoF) servo motor system, and the nonlinear control algorithm. Additionally, we explore the novel algorithm developed for path planning in the horizontal plane, considering the asynchronous angle adjustment between the leader and follower drones.

[0125] Figures 6A to 6C shows the different parameters of the system that enable control of the (3 DoF) servo motor system, whilst Figures 11A to 11C indicate the pitch, roll, and yaw coordinate reference frames. The SBT movement is highlighted for the maximum deviation in all three axes. The coordinate frame provides feedback for the areas under maximum pressure with a time stamp, which can be translated to estimate the angle of the SBT.

[0126] A mathematical equation in the Laplace domain typically represents the transfer function of a system. The transfer function relates the Laplace transform of the system’s output to the Laplace transform of its input. The single servo unit transfer function TS1(S) represents the dynamics between the servo motor input voltage and the resulting load angle. TS1(S) transfer function describes the servo load gear’s angle and the SBT dynamics’ location. Because this is a decoupled model, the x-axis servo does not affect the y-axis or z-axis response.

[0127] P(S), R(S), and Y(S) are the measured load positions in the pitch, roll, and yaw directions, respectively. In the Cartesian Coordinate system, ϕ(S) and θ(S) are the load anglesfor the shaft to rotate for the x-axis and y-axis motors, and Vm1(S), Vm2(S) Vm3(S) are the respective input voltages generated to give the respective load angles.

[0128] The open-loop system of the servo motor system for the SBT can be represented by the total transfer function T (S) Eq.1 derives the transfer function for the 1DoF for the SBT. T1(S) = TB1(S) · TS1(S) (1)

[0129] Where TB1(S) is the transfer function for the angle of the servo to the position of the SBT, and TS1(S) is the transfer function for the servo angle to the voltage. TB1(S) = P(S) / ϕ(S) (2) TS1(S) = ϕ(S) / Vm1(S) (3)

[0130] The two tilt angles [ϕ, θ] correspond to the movement of the servos in the system (αS1, αS2, αS3). The relationship between drone angle (input) and servo output is important for drone control systems. This relationship determines how the desired drone angle translates to the corresponding movement of the servo motor. The 3-degree of freedom test rack shown in Figures 10A and 10B, validates and fine-tunes this relationship. The test rack provides a controlled environment where the drone’s angle can be precisely adjusted, and the corresponding servo output can be measured and analyzed. This process helps ensure accurate and reliable control of the drones.

[0131] The estimated range movement of the pitch servo S1 tilt angle ϕ is from 0º to 40º, the roll servo S2 tilt angle θ ranges from 0º to 30º, and the yaw servo S3 is limited to 0º to 40º. The constants of the motor angles are represented by (f, t, r) and relate to the tilt angles using the kinematic equations defined in Equations (4)-(6).Where: h is the height of the SBT above the drone’s motors when it is parallel to the base, and|^^1 ^^2 |, | ^^2 ^^3 |, | ^^3 ^^1 | are the absolute distances between the SBT endpositions of servos S1 and S2, and motor arms S2 and S3, respectively.

[0132] These equations represent the SBT system’s final servo loop system function.

[0133] Incorporating asynchronous angle adjustments in the proposed path planning algorithm for drone payload delivery enables a deep analysis of the drone's movement in the x and y axes. The algorithm ensures that the drones can still move in a straight line in both the x and y directions by allowing for a maximum allowable difference in angles between the leader and follower drones. The path planning algorithm implemented in the horizontal plane is used to compute and resolve the effect of the asynchronous angle between the leader and follower. This algorithm considers the equation of motion for the rotational movement of the drone, considering the pitch angle (ϕ), roll angle (θ), and yaw angle (ψ).

[0134] The algorithm effectively controls the drone's rotational movements in the horizontal plane by transforming the angular body rates[ ^^^^] ^^to Euler rates[^^̇ ^^̇ ^^̇] ^^and using the calculated angular velocity in Eq. (7).

[0135] A gyroscope measures the angular velocities in the drone's x, y, and z axes. These measurements can be used to determine the shift in angular angles on the body frame. Using the inverse of Eq. (7) shown in Eq. (8), it is possible to gauge angular displacement based on the gyroscope readings. This equation allows calculation and tracking of the angular angle changes throughout the drone's movement.

[0136] When the angles between two drones differ significantly, it can lead to undesired consequences such as collisions or imbalanced forces that cause one drone to push or pull the other. To illustrate this, consideration can be given to the coordinate frames for each drone. Figure 7 shows the coordinate frames for the drones and the SBT. [ ^^1^^, ^^1^^, ^^1^^],and [ ^^2^^, ^^2^^, ^^2^^] represent drones 1 and 2, respectively. The combined coordinate frame, which represents the overall system, is denoted as [ ^^ ^^^^, ^^ ^^^^, ^^ ^^^^].

[0137] As shown in Eq.8, trigonometric operations compute changes in position and angle for every movement. The equation is a 3x3 matrix for each axis (x, y, and z). Eq.7 and 8 highlights the quadcopter frame’s rotation along each axis. The following Eq 9 is the matrix result of the sum between the two drone matrices (Dtx, Dty, and Dtz). Dttotal(t) = D1(t) + D2(t) (9)

[0138] Here Dttotal(t) is the product of the resulting angles for the drones shown in Eq. 10. Individual rotational movements of each of the resultant pitch angles (ϕ), roll angle (θ), and yaw angle (ψ) are the sum of the two drones. Dttotal(t) = ϕtotal(t) θtotal(t) ψtotal(t) (10)

[0139] Based on the Euler formula, the attitude of each drone calculates the location and direction of the net drone’s movement. Four rotors for each drone, the rotational speed, weight, and gravitational pull, compute the resultant angle of Yaw, Pitch, and Roll movement (not considering the external disturbance, currently). Equation (11) represents the consequentrotational angle for the swarm lift for the SBT. Here, ‘c’ is the cosine value of the angle and ‘s’ is the sine value.(11)

[0140] Several issues can arise if the two drones are connected directly without implementing the 3 DoF system. Firstly, achieving perfect mimicry between the two drones regarding time, angle, and acceleration becomes highly challenging. Any slight differences in the drones' responses and movements can lead to instability and erratic behaviour. Vibrations can become substantial without the 3 DoF system. The lack of separate control over each drone's pitch, roll, and yaw angles might result in unexpected oscillations and disruptions in the overall system. These vibrations can severely impact the payload's stability, potentially resulting in damage or loss during transportation.

[0141] Experimental results and observations are discussed below, and these provide further insights into the consequences of directly connecting the two drones without the 3 DoF system, highlighting the challenges and limitations that arise in terms of synchronization, stability, and vibration control. SBT Stabilization Mode

[0142] The SBT control model also incorporates two resistive panels on the tray to ensure high- resolution shifting and positioning measurements of the payload. The resistive panel principle has a gap separating top and bottom transparent conductive sheets with uniform resistance values. The contacted location in the X and Y directions is determined using the analog outputs. The SBT employs a grid system to precisely monitor the tray’s position and the target object’s alignment. The resistive panel records the shift in payload caused by the drone’s movements with a corresponding angle value.

[0143] Figure 8A shows the grid arrangement for both resistive panels used in the SBT. Each resistive panel is 1750x1350 mm in size. Every block in the grid of 437.5x450 mm will represent the payload's position. The leader drone computes the grid block position feedbacksent by the follower drone microcontroller to estimate the optimum position of the SBT. Additionally, the maximum deviation in the actuator system is observed for pitch and roll movements based on feedback from the resistive panel.

[0144] Table 2 provides the maximum deviation in the 3 DoF servo for pitch and roll movements based on the feedback from the resistive panel. Points A1-A3 and D1-D3 trigger the movement in the forward-backward direction to balance the pitch axis. The C1, C2, B1, and B2 issue right-left movements that balance the roll axis. Table 2: SBT Movement angle between 3-DoF Servo with 1 drone

[0145] The heat map in Figure 8B highlights the centre position as the lightest colour to determine the coordinates for the centre of the SBT. The lighter shades represent the maximum pitch value deviation for the SBT, and the darker highlights the zone for movement in roll angles.

[0146] Eq. 12 computes the respective motor angles for each legend pitch, roll, and yaw coordinates. In this equation, the P, R, and Y are the SBT coordinates, ‘C’ is the distance from the centre of the SBT to the servo motor (270 mm), ‘h’ is the distance from the drone frame to the SBT (100 mm), and ‘l’ is the distance between the 3DoF base and the SBT (60 mm).

[0147] A proportional–integral–derivative (PID) controller is a controller class that calculates the control sequence that must be implemented to stabilize the servo system to support the SBT between two drones and optimize the system’s future behaviour. Feedback is provided to a Zeigler-Nichols PID controller on the onboard microcontroller (distance between the SBT’s pose, payload’s current position and the drone’s centre) using the drone’s IMU alignment and PID gains to create input to the system.

[0148] Increasing KP steadily reveals optimal PID gains of KP = 7.7, Ki = 0.055, and Kd = 3.5. All experiments described below employ these gains. In Figures 8C to 8E, SBT performance in pitch (ϕ), roll (θ), and yaw (ψ) positions are depicted alongside error predictions. The range movement of the pitch servo S1 tilt angle ϕ is from 0º to 40º, the roll servo S2 tilt angle θ ranges from 0º to 30º, and the yaw servo S3 is limited to 0º to 40º.

[0149] Accordingly, the SBT balances the payload with more accuracy due to the position feedback from the resistive panel. The grid provides additional localization, the information can be further used to compute the shape of the payload. RESULTS AND DISCUSSION

[0150] The section discusses the experiments conducted to perform balanced lifts using the SBT between two drones. The 3 DoF servo configures the SBT position by optimizing the maximum deviation in all axes. This model determines the payload's position and adjusts the SBT to the optimal position for a balanced lift. Additionally, the approach incorporates asynchronous path planning techniques to verify the maximum threshold for drone flight deviation to maintain the flight path. The findings highlight the potential of aerial manipulation using innovative drone design and balancing techniques, improving drone navigation. Drone Angle and Balancing Position Optimization

[0151] The relationship between drone angle (input) and servo output is important for the control system of drones. This relationship determines how the desired drone angle translates to the corresponding movement of the servo motor. The 3-degree test rack shown in Figures 10A and 10B validates and fine-tunes this relationship. The test rack provides a controlled environment where the drone's angle can be precisely adjusted, and the corresponding servooutput can be measured and analyzed. This process helps ensure accurate and reliable control of the drones.

[0152] Figures 11A, 11B and 11C indicate the pitch, roll and yaw coordinate reference frames. The SBT movement is highlighted for the maximum deviation in all three axes.

[0153] Table 3 below provides a reference to 3 DoF servo angle comparisons with reference to the drone angles. (YPR)i gives the angles of the test rack. The data from drones 1 and 2 are similar because they share the same code and configuration. This similarity ensures that both drones operate synchronously, allowing for coordinated movements. Additionally, the fact that the drones are attached further reinforces their correlation and synchronized behavior. The average angle error values indicate the average discrepancy between the desired drone and achieved angles during the testing process. Table 3: Comparison angle between 3-DoF Servo with drone

[0154] The average angle error for yaw, pitch, and roll are 1°, 0.625°, and 2.6°, respectively, indicating the average deviation value that the achieved angle may deviate from the desired angle. These average angle errors provide information about the control system's performance and accuracy. They can assist in identifying areas for development and fine-tuning to reduce angle deviations and improve the overall control precision of the drones. Balancing Target Validation

[0155] Figures 12A and 12B show the leader-follower drone mid-flight lifting a partially filled bottle using the SBT. As depicted, the rotational angle for both drones is similar in Figure 12A, which is 45º, and in Figure 12B is 15º. The experiment results indicate that when both drones' pitch, roll, and yaw angles are the same, they move in the same direction; this implies that the synchronized control of the drones allows them to maintain coordinated movement.

[0156] The asynchronous angle behavior can be observed in Figures 12A and 12B, showcasing the drones' response and adjustment to the asynchronous angles, highlighting how they adapt and maintain their flight characteristics despite the differing angles. Further Results

[0157] Figure 13 compares measured servo angles with the actual drone angles experienced in the test rack shown in Figures 10A and 10B, to calibrate errors in the system. The data from drones are the same, ensuring that both drones operate synchronously, with a net movement angle being displayed on the y-axis. The scatter plots indicate the average angle error values between the desired drone and achieved attitude angles during testing. The points in Figure 13 highlight the maximum allowable yaw, pitch, and roll angle errors for a drone based on the drone’s input angle. For drone attitude at 40°, the corresponding average yaw angle is 35°. For the drone attitude deviations greater than 30°, the 3-DEE yaw shows an average error of 6°. However, for the drone attitude within the range of 30°, the corresponding average yaw angle error is less than 2°. Thus, the maximum allowable yaw angle is ±30°.

[0158] Similarly, the maximum pitch angle error is 3°, and the maximum roll angle error is 5° for input angles above 40°. Conversely, for input angles below 40°, the roll angle error should be less than 1°. These low deviation values emphasize the 3 DoF control for the drones that maintain precise control and stability of the drone during flight.

[0159] The relationship between the drone’s angle (input) and 3 DoF control can assist with control of the drones. The two drones were flown in different pitch and roll configurations attached to the SBT and the resultant angles were evaluated in Figures 14A and 14B and Figure 15. Figures 14A and 14B show the 3 DoF motion range results of angular values for roll and pitch based on variations of the asynchronous drone’s position. The test results highlight the net movement of the SBT when the angle of each drone is varied. In particular, Figure 14A shows the pitch angle for individual drones represented by ϕD1 for drone 1 and ϕD2 for drone 2, and Figure 14B shows roll values θD1for drone 1 and θD2for drone 2. In an asynchronous setting, the two quadcopters are tested with changing pitch and roll angles to verify the total cooperative behaviour of the system. It is also important to note that both the roll and pitch values are in the same direction for both drones.

[0160] Figure 15 shows the test results highlighting the net movement of the system when the angle of each drone is varied. However, in this case, only drone 1 angle values vary, i.e., D1 = 0, 5, 10, 15, while for drone 2, D2 = 5, 10, 15. The resulting pitch and roll values are shown. It is observed that the approximate resultant sum Dttotal(t) output is a summation of the individual drones’ roll and pitch values. For instance, D1 = 15º and D2 = 5º outputs a value of 20º, with a range of values between 18º and 22º.

[0161] It is also observed that the higher the difference between the angular values, the higher the error rate for the total Dttotal (t). Also, changing only pitch values affects the roll angle of the drone with slightly lower variation, and vice versa for changing only roll values. For instance, for roll variations as D1 = 15º and D1 = 0º, the observed roll is 15º, while the pitch angle is 13º. The average angle error for yaw, pitch, and roll are 1º, 0.625º and 2.6º, respectively.

[0162] These values indicate the average deviation from the desired angle. Average angle errors provide insights into the control system’s performance and accuracy, helping identify areas for development and fine-tuning to reduce angle deviations and improve the overall control precision of the drones.

[0163] Directly connecting two drones without the 3 DoF servo system risks instability due to response differences and vibrations, potentially harming payload stability. An experiment comparing rigid body attachment without adaptive control Figure 9B to servo-based adaptivecontrol shown in Figure 9A, using fixed and 3-DEE-equipped attachments respectively, showed excessive vibrations in the former. This illustrates the drone’s altitude hold mode at 2 meters with roll (ϕ), pitch (θ), and yaw (ψ) values over 8 seconds. Vibration (Vib) data from the MPU6000 on the Pixhawk flight controller, presented as RMS values for each axis (X, Y , Z) in m / s2within a ±16g and 40 Hz range, met typical RMS vibration thresholds: X-axis ≤ 2.2 m / s, Z-axis ≤ 2.8 m / s2, Y -axis ≤ 2.8 m / s2, ensuring stable flight. Implementing the 3-DEE system reduced vibrations across all axes, significantly enhancing overall stability and payload delivery performance. SBT Payload Balance

[0164] The impact of payload size on 3 DoF servo mobility and load distribution across the SBT grid was analysed. A heatmap representing servo system angle displacements across two 6x4 actuator grids on the SBT is shown in Figures 16A to 16D. Actuators in rows A and D control forward and backward movements, with rows B and C managing lateral roll movements. The values in the heatmap shows servo angle deviations.

[0165] The heatmap analysis under different payload conditions reveals distinct actuation patterns: In the no-payload condition of Figure 16B, servo angles in row A range from 14º to 34º, and in row D from -29º to -14º, reflecting the maximal range without load. With a small payload (4x2.5 cm) in Figure 16C, the range narrows, with angles from 16º to 27º in row A and -32º to -15º in row D. Under a large payload (6x6 cm) in Figure 16D, the range reduces further to 5º to 10º in row A and -11º to -7º in row D, indicating significantly restricted actuator movement due to increased load. This latter scenario vividly illustrates the substantially reduced actuator mobility due to the added load.

[0166] Throughout the three conditions, the outermost grid positions (A1-A3 and D1-D3) maintain a broader range of motion than the central servos (B1-C3), illustrating a spatially- dependent load distribution. This differential load effect across the grid demonstrates the 3- DEE performance and ability with load variations.

[0167] Payloads were tested on drones and with their displacement offset from the centre of the SBT during flight being measured. The servo system measured the payload position offset on the SBT during straight trajectory flights. The average time for servo position correctionand heat maps of servo angle deviations were then analysed to evaluate efficiency under different payload conditions. Results are summarised in Table 4. Table 4

[0168] The system demonstrates superior precision for the no-payload scenario with an average positional error of just 0.693 cm and a swift average completion time of 0.2 seconds to correct the payload offset from the centre. The results indicate high control accuracy and responsiveness without additional weight. With the introduction of a small payload, positional error slightly increases to 1.107 cm, accompanied by a longer completion time of 1.9 seconds.

[0169] This suggests a proportional relationship between payload size and control difficulty, with increased mass causing a modest drop in system precision and agility. The large payload condition further underscores this relationship, with the average positional error expanding to 2.32 cm and the completion time extending to 3.2 seconds.

[0170] The heat maps and tabulated data collectively suggest that the adaptive SBT control system maintains payload centring with high accuracy. In the depicted Figure 17 bar graph, movement characteristics of a drone are presented quantified in degrees, segregated into pitch, roll, and yaw across various positions. The bars indicate the individual degrees for pitch, roll, and yaw at drone positions ranging from -40º to 40º. Overlaid is a mean average line, illustrating the central tendency of movement across the three examined axes.

[0171] Data reveals a linear relationship between drone positions and their angular movements across three dimensions, underscoring high stability and precise control. Numerically, the SBT’s pitch, roll, and yaw are tightly clustered around -19º, -19º, and -18º respectively when the drone’s attitude is at -20º. This indicates a control system precision within approximately 1.6% of the mean value, vital for precision-demanding tasks. Furthermore, centre of mass imbalances are corrected swiftly using SBT feedback to recenter the payload, enhancing stability. It is noted that initially, the load was at the end of the pitch and roll movement, which would cause destabilization, and using the SBT feedback to resolve the balance of the bottleand the drones. For instance, feedback-driven adjustments allowed the SBT to stabilize the payload’s shift within 2.4 seconds during flight, as validated through observation. Dynamic Push-lift Stabilization

[0172] Experiments were conducted to assess the precision of the servo system to ensure the cooperative drones (D1 and D2) fly with the same attitude. During flight tests, the two drones have different attitudes while navigating a straight path, where D1’s pitch or roll was defined, and then we observe D2 converge to the same attitude during flight.

[0173] In the 3D line plot presented in Figure 18 it is observed that average movement of the pitch and roll of a cooperative system for 20 iterations of flight tests, demonstrating synchronous adjustment for change in individual drone movements. Each trajectory illustrates the evolution of movement averages as D2’s attitude increments while holding D1 at a constant. The solid lines trace the pitch, while the dotted lines represent the roll, facilitating a comparative analysis between the two types of movement.

[0174] For each fixed D1 attitude (0, 5, 10, 15), the mean average values for pitch and roll are depicted by ‘x’ markers. For instance, at a D1=10, the average pitch for the cooperative system is measured at 19.45º, demonstrating D2’s pitch attitude at an average of 10º. This suggests a proportional relationship between the two drones (D1 and D2) attitude for pitch and roll movements, indicative of the 3-DEE system’s responsive behaviour. Notably, the largest mean average for cooperative pitch is found at D1 = 15º with a value of 29.18º, closely followed by the cooperative roll’s mean average of 30.18º, indicating that D2 also converges to D2 = 15º on an average for pitch and roll. CONCLUSION

[0175] In conclusion, this above experimental validation demonstrates the approach can address the challenges faced in payload deliveries using drone swarms.

[0176] The self-balancing tray (SBT) can be provided as an add-on for drone swarms and used in conjunction with a path-planning algorithm for multi-agent drones tackled the limitations of payload capacity and unstable transportation for fragile or liquid materials.

[0177] The SBT can be integrated with a three Degrees of Freedom (DoF) servo control system with an average error rate of approximately 1 degree and a resistive plate distribution sensor to offer a high degree of stability and maneuverability for carrying payloads.

[0178] The nonlinear control system and asynchronous angle adjustment can be used to ensure precise control of the drones' movements in the horizontal plane, minimizing vibrations and maintaining object stability at 3-5 m / s flight speed. By addressing the limitations of payload capacity and instability, this research paves the way for advancements in drone swarm technology, opening up new possibilities for industrial applications.

[0179] Throughout this specification and claims which follow, unless the context requires otherwise, the word “comprise”, and variations such as “comprises” or “comprising”, will be understood to imply the inclusion of a stated integer or group of integers or steps but not the exclusion of any other integer or group of integers. As used herein and unless otherwise stated, the term "approximately" means ±20%.

[0180] Persons skilled in the art will appreciate that numerous variations and modifications will become apparent. All such variations and modifications which become apparent to persons skilled in the art, should be considered to fall within the spirit and scope that the invention broadly appearing before described.

Claims

THE CLAIMS DEFINING THE INVENTION ARE AS FOLLOWS: 1) A payload system for carrying a payload, the system including: a) a platform configured to support the payload; b) a payload distribution sensor configured to sense the distribution of the payload on the platform; c) at least two autonomous aerial vehicles; d) at least two coupling arrangements, each coupling arrangement being configured to couple a respective one of the vehicles to the platform; and, e) one or more processing devices configured to: i) determine a payload distribution in accordance with signals from the payload distribution sensor; and, ii) control operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted. 2) The payload system according to claim 1, wherein the coupling arrangements are configured to allow an orientation of each vehicle relative to the platform to be adjusted. 3) The payload system according to claim 2, wherein the orientation includes the pitch, yaw and roll of each vehicle relative to the platform. 4) The payload system according to claim 2 or claim 3, wherein each coupling arrangement includes actuators for controlling an orientation of the respective vehicle relative to the platform and wherein the one or more processing devices are configured to adjust the orientation of each vehicle in accordance with the determined payload distribution. 5) The payload system according to claim 4, wherein each coupling arrangement includes three actuators for controlling a pitch, yaw and roll of the respective vehicle. 6) The payload system according to claim 4 or claim 5, wherein the actuators include rotational servo motors. 7) The payload system according to any one of the claims 1 to 6, wherein each vehicle includes an altitude sensor, and wherein the one or more processing devices are configured to: i) determine vehicle altitudes in accordance with signals from the altitude sensors; and, ii) control operation of the vehicles in accordance with the determined altitudes. 8) The payload system according to claim 7, wherein each altitude sensor includes at least one of:a) a downward facing ranging device; and, b) a downward facing LiDAR. 9) The payload system according to any one of the claims 1 to 8, wherein the payload distribution sensor includes a resistive panel. 10) The payload system according to claim 9, wherein the payload distribution sensor includes orthogonally arranged sensing elements. 11) The payload system according to any one of the claims 1 to 10, wherein each coupling arrangement is magnetically attached to a respective vehicle. 12) The payload system according to any one of the claims 1 to 11, wherein the platform is situated above a centre of lift of each vehicle. 13) The payload system according to any one of the claims 1 to 12, wherein the one or more processing devices are configured to control the vehicles while carrying the load so as to at least one of: a) control an orientation of the platform; b) maintain the platform in a balanced position; c) maintain the platform substantially level while carrying the payload; and, d) maintain forces on the load substantially normal to a plane of the platform. 14) The payload system according to any one of the claims 1 to 13, wherein the one or more processing devices are configured to: a) determine a destination location; and, b) control the vehicles to transport the platform and payload to the destination location. 15) The payload system according to any one of the claims 1 to 14, wherein the one or more processing devices interact with a flight controller of at least one of the vehicles. 16) The payload system according to any one of the claims 1 to 15, wherein the one of the vehicles is a leader vehicle and other ones of the at least two vehicles are follower vehicles. 17) The payload system according to any one of the claims 1 to 16, wherein the system is configured to implement a path planning algorithm to determine a path to the destination. 18) The payload system according to claim 17, wherein the path planning algorithm incorporates asynchronous angle adjustments to resolve the effect of asynchronous angles between the leader and follower vehicles. 19) A method for carrying a payload, the method including: a) providing:i) a platform configured to support the payload; ii) a payload distribution sensor configured to sense the distribution of the payload on the platform; iii) at least two autonomous aerial vehicles; and, iv) at least two coupling arrangements, each coupling arrangement being configured to couple a respective one of the vehicles to the platform; and, b) in one or more processing devices: i) determining a payload distribution in accordance with signals from the payload distribution sensor; and, ii) controlling operation of the vehicles in accordance with the determined payload distribution to thereby allow the platform and payload to be lifted.

Images