Method for saving energy consumption of a UAV and improving its hovering accuracy

Abstract

A multi-rotor unmanned aerial vehicle ("UAV") is provided, including methods of operating it so that it can perch on a surface in an environment without additional equipment, thereby saving energy. The UAV has a frame that supports at least four variable speed reversible motors spaced around the frame. Each motor drives a propeller that is arranged to hover the frame. Sensors are provided on the UAV for detecting a distance between the UAV and a target surface. A computer is provided for executing software modules to control operation of the motors to cause the UAV to perform flight maneuvers including at least: (a) approaching the target surface, (b) using a "ceiling effect" to contact the target surface so as to perch on the surface and reduce energy consumption of the UAV, and (c) disengaging from the target surface.

Description

[0001] This international patent application claims the benefit of U.S. Provisional Patent Application Serial No. 63 / 018,982, filed May 1, 2020, the entire contents of which are incorporated herein by reference. TECHNICAL FIELD

[0002] The present invention relates generally to reducing the energy consumption and improving the hovering accuracy of a multi-copter unmanned aerial vehicle (“UAV”) by docking it on a planar structure (including but not limited to a ceiling, a wall, and the ground) on-site. BACKGROUND

[0003] In recent years, there has been an increasing demand in the construction industry and infrastructure management to deploy UAVs in order to improve task efficiency, reduce labor costs, and enhance site safety. The unique and superior maneuverability of UAVs in 3D space is believed to enable many applications that are not possible with ground vehicles or are too dangerous for humans. For example, using UAVs for automated construction site monitoring would enhance site safety by identifying potential risks. A UAV equipped with a laser scanner can automatically acquire a 3D digital model of an unknown environment, which can be applied to building information modeling (BIM) to improve task efficiency and enable routine modeling of large multi-story buildings. Other applications, such as infrastructure inspection (e.g., bridges, towers) and emergency transport (e.g., first-aid medicine, blood samples), also benefit greatly from the capabilities of UAVs.

[0004] Despite these exciting possibilities, the limited working time of multi-copter UAVs due to the continuous propeller rotation and the resulting energy consumption severely hinders their practical use in long-term tasks, such as construction site monitoring and modeling [1, 2]. Figure 1 The working time of different sizes of commercial off-the-shelf (“COTS”) multi-copter UAVs manufactured by DJI Technology Co. Ltd. was summarized based on the momentum theory disclosed in

[56] , i.e., W b ·D·W t -3 / 2 (m / √kg) prediction. In the equations and figures, W b is the battery weight, W t is the total weight, D is the propeller size, and N is the number of rotors. The typical working time of a multi-copter UAV is a few tens of minutes. When carrying additional payloads (e.g., a laser scanner for BIM) or being smaller in size (e.g., to have better maneuverability in a crowded construction site), the working time continues to decrease, as shown in Figure 1 These short working times result in the batteries driving the propeller motors needing to be charged frequently, which greatly limits task efficiency and increases facility costs.

[0005] Birds and insects perch or rest on natural objects (such as tree branches, power lines) to conserve energy. In addition, perching allows them to obtain a vantage point for observation (perching and gazing)[3]. Kingbirds perch on power lines to obtain an unobstructed view of open spaces, which is advantageous for them to scout for potential prey. Many other birds (such as rock pigeons) perch on tree branches or high power lines to conserve energy while being far away from potential predators (see Figure 2A ) Inspired by nature, researchers have started to employ similar perching behavior in UAVs to extend their working time[4]. Depending on the applicable perching structure, these developments fall into three categories: wall, ceiling, and branch-like shapes.

[0006] Perching on walls. Anderson et al.[5] used adhesive pads with liquid glue to perch a light-weight aircraft on a vertical surface. Separation of the plane from the surface was achieved by cutting the pad with a razor blade. In the work of Mellinger et al. and Bayraktar et al.[6, 7], magic tape was used to attach a UAV to an inclined surface. By mimicking perching with claws, Kovac et al.[8] developed a perching mechanism with two needles that can pierce into a wall. Asbect et al. and Roderick et al.[9]

[14] described the use of microspikes to engage with roughness on a wall Figure 2B ). Both methods typically have minimal force support and are demonstrated only on light-weight flying machine insects or aerobatic aircraft. For heavier multi-copter UAVs, gecko-inspired dry-adhesive grippers are used to provide stronger adhesion[15-18]. Other grippers, such as fiber-adhesive pads

[19] and dry-adhesive pads

[20] , and control methods[21, 22] are also being actively researched. In all the above methods, the UAV has to turn the bottom side forward to perch on a wall. See Figure 2C , which illustrates this action using gecko-inspired grippers as in

[18] . This procedure typically blocks the field of view of the task sensor (e.g., camera 14), which is also mounted on the bottom side of the vehicle to enable video acquisition in operation. The installation of these grippers can also cause mechanical interference to the sensor.

[0007] Perching on ceilings. Compared to perching on walls, there is much less research on perching on ceilings. Roberts et al.

[23] used a ring-shaped magnet to perch a quadcopter UAV on an iron ceiling. Graule et al.

[24] proposed a switchable electro-adhesive that enables controlled perching and detachment on various materials: glass, wood, and natural leaves. This adhesion mechanism requires a 1000V voltage supply and generates a small electrostatic force that can only lift a machine insect weighing less than one gram. See Figure 2DSanchez-Cuevas et al.

[25] utilized the ceiling effect [26, 27] of a single propeller to perch a normal-sized quadcopter on a bridge. Various experiments have been conducted on single propeller and full quadcopter vehicles to show energy savings versus propeller-to-ceiling distance. More recently, Chirarattananon et al.

[28] studied the ceiling effect using a combination of momentum theory and blade element method and demonstrated a conceptualized perching of a light-weight micro-sized quadcopter. See https: / / www.bitcraze.io / crazyflie-2 / .

[0008] The ceiling effect can be described as an effect where the upstream air from a propeller approaching a flat ceiling from below will interact with the ceiling. The result is that a propeller rotating at a certain speed (i.e. requiring a certain power) will produce more thrust when closer to the ceiling. It can be imagined that the "extra thrust" is generated by the propeller air upstream-ceiling interaction without consuming further power from the motor. In other words, when providing the same thrust (e.g. UAV weight), the propeller will consume less power when approaching the ceiling than in the air. With careful design, the power is typically reduced by 30% or even more.

[0009] Perching on tree branches. For tree-like structures, researchers [29-32] have developed passive gripping mechanisms by mimicking the feet of songbirds. This mechanism utilizes the weight of the UAV to passively actuate the grippers (see Figure 2E ). Other researchers [33-35] have disclosed similar grippers for perching on tree-like structures, where the grippers are actuated by servo motors, allowing stable and controllable perching and disengagement. In both cases, the grippers are mounted on the underside of the UAV, which blocks the view of the sensors and can cause mission interference or even suspension.

[0010] Existing gaps. In short, while these methods work well in specific cases, none of them can both perch on walls and ceilings, which is a widespread scenario in real-world applications such as as-built BIM. Moreover, most of them require grippers, which complicate the mechanical design and interfere with the mission. As a result, none of these designs has been widely adopted by practical multicopter UAVs. To improve the efficiency and accuracy of practical UAVs, it is crucial and urgent to propose a new perching methodology that is compatible with existing UAVs, requires minimal or even no mechanical accessories, and does not interfere with the mission.

[0011] One of the inventors, Fu Zhang, has done a lot of work on UAV design

[36] , implementation [37-39], navigation [40-42] and control [43-47]. Most notably, in work reported in a cover article in Science Robotics

[48] , Fu Zhang developed a claw-like deformable landing gear that enables a UAV to perch on various natural structures, such as tree branches, utility poles and rooftops Figure 2F ), but not on planar structures. All of this work has been implemented and tested on real systems and will greatly benefit the development and control of the UAV of the present invention. SUMMARY

[0012] The present invention aims to increase the working duration and precision of a multi-copter UAV by exploiting the ceiling effect, exploiting the contact between the UAV and planar structures in the field (including walls, ceilings and the ground). The present invention also helps to prevent the UAV from crashing in bad weather conditions (e.g., strong winds) by perching on nearby buildings. The present invention also enables the UAV to land without the landing gear blocking the view of the on-board sensors.

[0013] One design involves perching on planar structures, while the other involves moving on the ground. For each design, trajectory generation and control algorithms are presented to achieve autonomous, fast, precise perching and air-ground hybrid movement. All designs and algorithms are shown integrated into a real quadcopter UAV, which is tested in real-world experiments. BRIEF DESCRIPTION OF DRAWINGS

[0014] The foregoing and other objects and advantages of the invention will become more apparent from the following detailed description and accompanying drawings, in which like reference numerals refer to like elements in the several views, and in which:

[0015] Figure 1 is a plot of UAV flight time versus prediction based on momentum theory, where battery weight, total weight, propeller size and number of rotors are parameters;

[0016] Figure 2A is a photo of a bird perched on a horizontal wire, Figure 2B is a photo of a microspine, Figure 2C is a photo of a gecko-inspired gripper; Figure 2D is a photo of an electro-viscous gripper, Figure 2E is a photo of a passive gripper, Figure 2F is a series of photos showing an active gripper;

[0017] Figure 3Ais a schematic view of a UAV according to the present invention, which utilizes the ceiling effect to perch on a planar structure, including walls, ceiling and floor, and which preserves the maximum possible field of view in all cases, Figure 3B is a schematic view of a UAV according to the present invention perched on a wall, ceiling and floor, Figure 3C is a cross-sectional view of a UAV of the present invention, which perches on a wall due to the static friction force that counteracts gravity, and which shows the forces acting on it;

[0018] Figure 4A shows a UAV of the present invention approaching, perching and disengaging a ceiling, Figure 4B shows a UAV of the present invention approaching, perching and disengaging a vertical wall, Figure 4C shows a UAV of the present invention approaching, perching and disengaging a floor;

[0019] Figure 5 is a block diagram of an exemplary embodiment of a cascaded control structure, which provides a compatible interface for all phases of the perching action; and

[0020] Figure 6A is a photograph of a UAV of the present invention perched on a ceiling, Figure 6B is a photograph of a UAV of the present invention perched on a wall, Figure 6C is a graph showing a UAV of the present invention with more than 30% power reduction. DETAILED DESCRIPTION

[0021] The present invention aims to generate Figure 3A The UAV 10 is shown in two orientations in Fig. 1. As Figure 3B shown, it is designed to perch on or connect to planar structures, including walls 20, 21 and ceiling 22 as well as flat floor 23, utilizing the "ceiling effect". This preserves the maximum possible field of view in all cases. When perched on a wall, static friction builds up and counteracts gravity, as Figure 3C shown.

[0022] The perching is performed using a standard quadcopter 10 Figure 3A ) with propeller guards 12 serving as a support structure when the vehicle contacts a planar structure, e.g. a wall. In contrast to prior designs disclosed in [6, 15-20, 23, 24, 29-35, 48], the perching method of the present invention does not require any additional components, such as grippers (propeller guards 12 are required anyway for safety purposes), and preserves the maximum possible field of view. Furthermore, the full and firm contact with the environment greatly improves the accuracy of the UAV.

[0023] This perching method is extended to a novel landing strategy, where the UAV is flipped upside down 10' and uses the propeller guards 12 as landing gear on the ground 23. Figure 3B This landing maneuver allows the on-board cameras 14A, 14B to have a large clear field of view, eliminating the need for retractable landing gear. This is particularly useful for UAVs with 360-degree on-board cameras, such as the DJI Phantom X (concept design available at https: / / youtu.be / ec1EF2UaQ4U). In particular, to not block the camera's view, traditional landing gear needs to be retracted after take-off. The retraction system adds more weight and complexity to the UAV, and is therefore disadvantageous. When perching on a ceiling, wall, slope, or ground, the present invention does not block the view of any on-board camera, and therefore does not require retractable landing gear.

[0024] According to the present invention, perching requires the UAV to make full contact with the environment, which presents various challenges for its control:

[0025] (1) The UAV must approach the wall in an unstable upright posture, which requires precise trajectory generation and tracking control;

[0026] (2) When approaching the environmental structure, the downstream propellers will be affected and cause considerable disturbance to the system;

[0027] (3) The controller needs to adapt to the unknown and varying static friction caused by contact; and

[0028] (4) When in contact with the environmental structure, the UAV dynamics suddenly switch, and the controller is required to respond correctly.

[0029] To address these challenges and enable autonomous, safe, precise, and fast perching, the system design is optimized by integrating trajectory generation, tracking control, and perching maneuvers.

[0030] To enhance the perching capability of the UAV, dampers 13 are located on the edges of each propeller guard 12, as shown in Figure 3AAs shown. The damper can (a) absorb kinetic energy upon impact with the ceiling, (b) increase friction with the wall during parking, and (c) improve compliance when contacting an uneven ceiling or wall. Furthermore, to enable fully controlled parking, the open-source electronic speed control (“ESC”) disclosed in

[49] can be used, which allows the propeller to rotate in reverse and generate both positive and negative thrust. Compared to the variable pitch propeller disclosed in

[50] , which also generates reversible thrust, this invention does not require a complex pitch-changing mechanism, but only requires an upgrade to the ESC firmware, thereby maximizing compatibility with existing UAVs. However, variable pitch propellers are also compatible with this invention, provided they can reverse propeller thrust.

[0031] like Figure 3A The UAV shown has four propellers driven by individual reversible motors. The motors are powered by batteries 15. The UAV is also equipped with a time-of-flight (“TOF”) laser ranging system 15 and passive omnidirectional wheels 19. The UAV is under the control of an onboard computer 16.

[0032] Preferably, an efficient trajectory generation algorithm is developed that allows a multi-rotor UAV to approach a planar structure at a desired speed and attitude. Reversible propellers in a UAV system present two novel challenges to trajectory generation that have been little studied in existing work disclosed in [51, 52]: (1) ambiguity in determining thrust input and attitude (e.g., simultaneously flipping the UAV and changing the sign of the thrust does not change the acceleration), and (2) degradation in trajectory tracking performance due to motor delay when the reversing motor rotates. This invention overcomes these challenges with a hierarchical trajectory generation framework by dividing the entire trajectory into smaller sub-trajectories (i.e., motion primitives) according to the actual task. For each sub-trajectory, all propellers maintain their rotational direction (as does the thrust), thus enabling the use of existing trajectory generation algorithms (e.g., as disclosed in

[52] ). Furthermore, this ensures that all propellers simultaneously reverse their rotation, and only between two sub-trajectories, thus preventing frequent reversal rotations when performing aggressive maneuvers (e.g., parking). A core function library was developed that computes the optimal sub-trajectory from any starting state to any given target state. Real-time trajectory generation was developed by utilizing the closed-form structure of the optimal solution and leveraging the availability of high-performance onboard computing equipment16 (e.g., Intel NUC boards). Information about this computer motherboard can be found at https: / / www.intel.com / content / www / us / en / products / boards-kits / nuc / boards / nuc7i7dnbe.html

[0033] A control framework was developed as a unified control framework that can be used for all phases of perching, which is a challenging task for contact-rich robots. The unique perching design of the invention allows the use of a cascaded control structure Figure 5 ), which is widely used in existing UAVs and maximizes the compatibility of the invention with other UAV systems. In this control structure, the outer loop controller tracks the desired trajectory, and the inner loop controller regulates the thrust input and angular velocity. This control structure provides a compatible interface for all phases of perching and supports easy switching between them. For example, when flying towards the perching location, the tracking controller 15 is necessary for accurate tracking of the planned trajectory so that the UAV reaches the perching location with the desired speed and attitude. During the contact establishment and maintenance phase, direct instructions for throttle and torque (or attitude) are desired.

[0034] Controller parameters can be optimized in the frequency domain by loop shaping techniques that can be found in [53, 54] developed in one of the inventor's previous works. In addition, additional feedback compensation (e.g., disturbance observer as disclosed in

[46] ) can be added in order to further attenuate various aerodynamic disturbances caused by the environment structure. Feedforward control, such as iterative learning control that learns from previous iterations and adaptive control that adjusts parameters in real-time as disclosed in

[55] , is used to improve control performance.

[0035] The power savings caused by the "ceiling effect" exploited by the invention when the UAV perches on the ceiling were determined. The results are shown in Figure 6C . Perching was performed by a human pilot carefully flying the UAV towards the ceiling. See Figure 3B and 6A . Once full contact was observed, the throttle was gradually decreased until disengagement occurred. The results show that the power reduction reached 30%. Although the power saved is not as much as the gripper-based solutions disclosed in [6, 15-20, 23, 24, 29-35, 48], this perching is achievable by existing UAVs without any additional devices.

[0036] Based on the developed trajectory generation and tracking control algorithms, the actions of the UAV were developed that enable autonomous, safe, and robust perching on the ceiling. Perching is divided into three phases: approach, contact establishment and maintenance, and disengagement Figure 4A . In the approach phase, the UAV moves from its current state to a target state in which the position is just below the ceiling, and both the velocity and acceleration are zero. The zero acceleration of the target means that the UAV is in a hover attitude with zero pitch and roll angles, which is necessary for safe perching. The approach trajectory is repeatedly generated by using the developed algorithm, in which the thrust is always positive.

[0037] Once the UAV is in the target position directly below the ceiling, contact will be established automatically with the help of the ceiling effect. Thereafter, the torque will be set to zero while gradually reducing the throttle to save energy. It is believed that the accuracy will improve when hovering on the ceiling and it will be possible to determine the minimum throttle that can be achieved at different propeller-to-ceiling distances.

[0038] The disengagement phase is the inverse of the approach phase: moving the UAV from the hovering position to a position well below the ceiling with zero speed and acceleration. This process is directly implemented through the trajectory generation and tracking control discussed above.

[0039] Figure 6B A hovering experiment of a quadrotor UAV on a wall with the ceiling effect is shown. The use of the invention on a vertical wall also applies to an inclined surface. The UAV has non-reversible propellers, but is still able to hover on the wall by a human pilot. In this context, the UAV first moves slowly towards the wall. Once the two propeller guards on one side contact the wall, they form a pivot and enable the UAV to rotate until the other two propeller guards also contact the wall. Similar to the ceiling, hovering on the wall saves 30% of the battery power.

[0040] According to the invention, an action enabling autonomous and aggressive hovering is developed by fully exploiting the effect of reversible thrust and friction during the simulation of contact. Similar to the ceiling, wall hovering is divided into three phases: approach, contact establishment and maintenance, and disengagement Figure 4B ). The whole process is actively controlled by the controller 16 and the trajectory is repeatedly computed once the UAV is in the air (i.e. approach and disengagement).

[0041] The approach trajectory consists of two sub-trajectories: the first sub-trajectory accelerates the UAV from the current state to an intermediate state, and the second sub-trajectory decelerates the UAV to the target state. The velocity of the intermediate state is parallel to n, the normal of the wall, and the acceleration is g - εn, which means the thrust is εn. A small thrust ε is used instead of zero thrust because zero thrust would make the attitude indeterminate (e.g. all attitudes result in the same zero acceleration). Since ε is very small, the propellers essentially rotate at idle RPM, i.e. the minimum speed at which the motors work normally, thus providing practically negligible thrust and allowing the rotation to be reversed for the second sub-trajectory. Similarly, in the hovering position, the velocity and acceleration are set to zero and g + εn, respectively. As a result, the first sub-trajectory always has a positive thrust, while the second sub-trajectory is negative, both of which can be generated by the trajectory algorithm in

[52] . Between the two sub-trajectories, all propellers reverse their direction, as does the thrust. The intermediate velocity and position parameterize the two sub-trajectories, and their values are determined by minimizing the total cost (e.g. jerk, energy).

[0042] Once the UAV is in the parked position, contact is established with positive throttle, which creates a pushing force against the wall. After this, static friction builds up and counteracts gravity. Assuming T u and T l are the thrusts of the upper and lower propellers, respectively, both positive, N u , N l are the normal forces, f u ; f l is the static friction force (see Figure 3C ), then the sufficient and necessary conditions to maintain contact are:

[0043] Zero horizontal forces: T u + T l = N u + N l (1a)

[0044] Zero vertical forces: f u + f l = mg (1b)

[0045] Zero pitch moment: l h (f u + f l ) = l v (T u - N u )- l v (T l - N l ) (1c)

[0046] Effective normal forces: N u ≥ 0; N l ≥ 0 (1d)

[0047] From (1a-1c), we can solve for the normal forces: N l = T l + (lh / lv)mg; N u = T μ - (lh / lv)mg. Since static friction f is related to normal N through f < μsN, where μ s is the static friction coefficient, (1a-1b) leads to μ s (T u + T l ) ≥ mg. Moreover, propeller power is related to thrust through PαT√T from the momentum theory disclosed in

[56] . In all these cases, the minimum power is obtained as follows.

[0048]

[0049] The optimal solution of equation (2) is obtained from the KKT conditions described in

[57] and is divided into three cases: (1) μ s ≤l v / l h , then T μ * = T l * = mg / 2μ s ; (2) l v / l h ≤ μ s ≤ 2(l v / l h ), then T μ * = (l h / l v )mg, T l * = mg[l / μ s - (l h / l v )]; (3) μ s ≥ 2(l v / l h ), then T μ * = (l h / l v )mg; T l * = 0. In practice, the coefficient μ s is usually unknown as it also depends on the material of the environment. To find the minimum thrust and power, an active search procedure is used as follows: after establishing contact, in the first case, the search is performed by setting the torque to zero (i.e., T μ = T l ) and then gradually decreasing the throttle (i.e., case 1). If a contact loss is detected (e.g., measured from an inertial measurement unit (“IMU”)), then T u is fixed, T l is gradually decreased (i.e., case 2). The search procedure terminates upon detecting another contact loss or T l reaches zero (i.e., case 3). It should be noted that both T μ and T l monotonically decrease with the coefficient μ s (total power does as well) even in the worst case (i.e., case 1), the minimum thrust is T μ * = T l * = mg / 2μ s , if μ s> 1, which is less than the thrust when hovering at the ceiling.

[0050] Furthermore, various damper materials such as rubber and dry adhesives can be explored to maximize the coefficient μ of common wall materials such as concrete, plaster, brick, and wood s Furthermore, the increase in precision of the UAV when hovering on a wall according to the present invention can also be determined experimentally.

[0051] The disengagement phase is achieved by setting the throttle to a negative value and the attitude to the hover attitude. This will trigger the counter-rotation of the upper propellers and the lowering or even counter-rotation of the lower propellers. The negative throttle will make the UAV quickly leave the wall, while the attitude command will make the UAV rotate to the hover attitude. A drop in altitude can occur during this process, but the possible drop in altitude is usually small because the disengagement is very fast. A recovery trajectory can also be used to guide the UAV back to the normal hover state after disengagement.

[0052] To achieve a hover on the ground, there should be a controllable flip maneuver that guarantees a safe landing. The flip is a stunt maneuver that requires precise sensing and control.

[0053] Previous work has been done with fixed-pitch propellers to perform 360° flips (see [58-60]) and with variable-pitch propellers to perform 180° flips (see

[50] ). However, 180° flips using reversible fixed-pitch propellers have only been demonstrated by skilled pilots on small quadcopters, without position control. Even then, unlike the present invention, previous drones could not land in that orientation. Reversible fixed-pitch propellers have a larger motor delay compared to variable-pitch propellers, which presents a significant challenge for the flip. However, the present invention is still applicable to variable-pitch propellers.

[0054] The present invention utilizes a three-phase process: ascent, flip, and descent (see Figure 4C ) In the first ascent phase, the UAV ascends at a speed v tand acceleration g + εe3 in the vertical direction to target height h. Small thrust εe3 means that the UAV is in hover attitude with all the propellers at idle RPM. In the second flipping phase, the throttle is kept at zero while the attitude is set to inverted hover attitude. This will trigger half of the propellers to reverse their rotation and start the flip. Also, zero throttle results in zero horizontal displacement during the flip. At the end of the flip, all the propellers will be at idle RPM with reversed rotation. In the third descending phase, the UAV will increase its propeller rotation (in the opposite direction) to slow itself down until it touches the ground. The trajectory is recalculated and tracked repeatedly during the ascent (positive thrust) and descent (negative thrust) so that no horizontal displacement is generated. Height h and velocity v t Adjusted by experimentation.

[0055] The system of the present invention is shown in Figure 3A with a single passive omni-wheel 19 to get force support from the ground, allowing the propellers to rotate at much lower speed (and thus lower power), which is necessary to maintain satisfactory control. When moving on the ground, the four propellers actuate the UAV in the same way as in the air, i.e. different thrust actuates attitude, which then actuates translation, thus allowing the use of the same control structure and trajectory generation method in different flight modes. The transition trajectory from air to ground is generated by (a) setting the target position to the point of contact with the ground, (b) setting the target velocity and acceleration to be parallel to the ground. Damper 13 is used to absorb the remaining kinetic energy when hitting the ground.

[0056] Moving on the ground also induces ground effect, which saves even more energy ( Figure 6C ) as disclosed in

[61] . The minimum achievable throttle, the power saved due to throttle reduction and ground effect, and the hybrid motion along the ground can be studied in a real UAV operating according to the present invention, in the air, on the ground, and during air-ground transitions.

[0057] All the developed algorithms are integrated into a complete quadrotor system that can automatically perform navigation, planar structure detection, trajectory generation, and tracking control. Figure 5 An exemplary embodiment of the control system is shown in. In Figure 3AIn the UAV of FIG. 1, the autopilot executes control algorithms, and the onboard microcomputer 50 executes real-time trajectory generation, autonomous navigation, and planar or structural detection. Sensors, such as the Intel RealSense d400 disclosed in

[62] , provide input so that the UAV can autonomously navigate, build a map of the environment, and detect planar structures therein. For navigation and map building, existing open-source software for RealSense can be used directly. See, SLAM with D435i, https: / / github.com / IntelRealSense / realsense-ros / wiki / SLAM-with-D435i. For planar detection, the methods described in

[63] are employed, which have been shown to be efficient and effective on real datasets. These methods are optimized for large vertical and horizontal planes.

[0058] The light-weight single-point time-of-flight (ToF) laser ranging module 15 disclosed in

[64] is used to measure distances to planar structures with millimeter-level precision, thus enabling precise hover control and reliable contact detection when used with the onboard IMU 16. With all of this, the hover and landing maneuvers can be combined with onboard sensing to form a complete and autonomous system.

[0059] Figure 5 The cascaded control structure of FIG. 1 is widely used in existing UAVs and maximizes the compatibility of the present invention with other UAV systems. In this control structure, the outer loop of the onboard computer 50, flight controller 52, and quadcopter UAV tracks a desired trajectory, while the inner loop controller of the onboard computer 50 regulates thrust inputs and angular velocities, which includes the visual-inertial navigation unit 60, local map building unit 62, planar detection unit 64, mode switching logic 66, contact detector 68, and trajectory generation 61. This control structure provides a compatible interface for all stages of hovering and supports easy switching between them. Sensors 84 on the quadcopter UAV 54 provide signals about the UAV’s performance to the onboard computer. These signals are received in the visual-inertial navigation unit 60 of the computer 50. The output of the unit 60 is used for local map building in the unit 62 and trajectory generation in the unit 61. The output of the local map building unit is received in the planar detection unit 64, which in turn provides information to the mode switching logic 66. Mode switching is set to depend on the desired maneuver, e.g., hovering on a ceiling, wall, or floor. Based on the information from the mode switcher 66 and the visual-inertial navigation unit 60, the trajectory generation unit 61 generates signals that are applied to the position controller 70 located in the flight controller 52. The position controller sets the thrust and attitude.

[0060] At the same time, the output of the mode switcher 66 is also input to the contact detection unit 68. Its output determines whether the attitude signal from the position controller is applied to the attitude controller 72. The output of the attitude controller is then applied to a proportional-integral-derivative controller ("PID") 75. The thrust output of the position controller 70 is applied to a PID 73. The output of the PID 73 is thus the throttle signal, and the output of the PID 75 is the motor torque signal. Based on the output of the contact detector 68, these signals are applied to a mixer 76.

[0061] The output of the mixer 76 is applied to the actuator dynamics system 80 of the UAV, which in turn drives the UAV dynamics system 82. The effect of the dynamics system or motor results in a change in the output of the sensor 84, thus completing the outer control loop.

[0062] When flying towards the parking position, the sensor 84 detects the surface or plane. This information is passed on to the plane detector 64. The approach to the surface is also passed on to the trajectory generator, which has stored the algorithm. The chosen trajectory depends on the mode signal from the mode switcher 66. This trajectory generation signal goes through the flight controller 52 to control the throttle and torque applied to the motor (dynamics system 82) of the UAV. In this way, the flight path of the UAV is controlled so that it precisely tracks the planned trajectory so that the UAV arrives at the parking position with the desired speed and attitude. When in contact with the surface, the contact detector 68 changes the flight controller 52 so that the UAV remains in contact with the surface due to the ceiling effect.

[0063] While the application has been particularly shown and described with reference to particular embodiments thereof, it will be understood by those skilled in the art that various changes in the form and details can be made therein without departing from the spirit and scope of the application. In particular, the foregoing detailed description has been presented for purposes of illustration and description only. It is not intended to be exhaustive or to limit the application to the precise form disclosed. The description makes no

[0064] List of References

[0065] The following references are cited in this application. Each such reference is incorporated herein by reference in its entirety.

[0066] [1] D. Floreano and R. J. Wood, "Science, technology and the future of small autonomous drones," Nature, vol. 521, no. 7553, p. 460, 2015.

[0067] [2] Y. Mulgaonkar, M. Whitzer, B. Morgan, C. M. Kroninger, A. M. Harrington, and V. Kumar, “Power and weight considerations in small, agile quadrotors,” in Micro- and Nanotechnology Sensors, Systems, and Applications VI, vol. 9083. International Society for Optics and Photonics, 2014, p. 90831Q.

[0068] [3] T. W. Danko, A. Kellas, and P. Y. Oh, “Robotic rotorcraft and perch-and-stare: Sensing landing zones and handling obscurants,” in ICAR’05. Proceedings, 12th International Conference on Advanced Robotics, 2005. IEEE, 2005, pp. 296-302.

[0069] [4] M. Kovac, “Learning from nature how to land aerial robots,” Science, vol. 352, no. 6288, pp. 895-896, 2016.

[0070] [5] M. Anderson, “The sticky-pad plane and other innovative concepts for perching uavs,” in 47th AIAA Aerospace Sciences Meeting Including The New Horizons Forum and Aerospace Exposition, 2009, p. 40.

[0071] [6] D. Mellinger, N. Michael, and V. Kumar, “Trajectory generation and control for precise aggressive maneuvers with quadrotors,” The International Journal of Robotics Research, vol. 31, no. 5, pp. 664-674, 2012.

[0072] [7] S. Bayraktar and E. Feron, “Experiments with small helicopter automated landings at unusual attitudes,” arXiv preprint arXiv:0709.1744, 2007.

[0073] [8] M. Kovac, J. Germann, C. Hurzeler, R. Y. Siegwart, and D. Floreano, “A perching mechanism for micro aerial vehicles,” Journal of Micro-Nano Mechatronics, vol. 5, no. 3-4, pp. 77-91, 2009.

[0074] [9] A. T. Asbeck, S. Kim, M. R. Cutkosky, W. R. Provancher, and M. Lanzetta, “Scaling hard vertical surfaces with compliant microspine arrays,” The International Journal of Robotics Research, vol. 25, no. 12, pp. 1165-1179, 2006.

[0075]

[11] A. Lussier Desbiens, A. T. Asbeck, and M. R. Cutkosky, “Landing, perching and taking off from vertical surfaces,” The International Journal of Robotics Research, vol. 30, no. 3, pp. 355-370, 2011.

[0076]

[12] A. L. Desbiens, A. T. Asbeck, and M. R. Cutkosky, “Scansorial landing and perching,” in Robotics Research. Springer, 2011, pp. 169-184.

[0077]

[13] D. Mehanovic, J. Bass, T. Courteau, D. Rancourt, and A. L. Desbiens, “Autonomous thrust-assisted perching of a fixed-wing uav on vertical surfaces,” in Conference on Biomimetic and Biohybrid Systems. Springer, 2017, pp. 302-314.

[0078]

[14] W. R. Roderick, H. Jiang, S. Wang, D. Lentink, and M. R. Cutkosky, “Bioinspired grippers for natural curved surface perching,” in Conference on Biomimetic and Biohybrid Systems. Springer, 2017, pp. 604-610.

[0079]

[15] E. W. Hawkes, D. L. Christensen, E. V. Eason, M. A. Estrada, M. Heverly, E. Hilgemann, H. Jiang, M. T. Pope, A. Parness, and M. R. Cutkosky, “Dynamic surface grasping with directional adhesion,” in 2013 IEEE / RSJ International Conference on Intelligent Robots and Systems. IEEE, 2013, pp. 5487-5493.

[0080]

[16] H. Jiang, M. T. Pope, E. W. Hawkes, D. L. Christensen, M. A. Estrada, A. Parlier, R. Tran, and M. R. Cutkosky, “Modeling the dynamics of perching with opposed-grip mechanisms,” in 2014 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2014, pp. 3102-3108.

[0081]

[17] H. Jiang, M. T. Pope, M. A. Estrada, B. Edwards, M. Cuson, E. W. Hawkes, and M. R. Cutkosky, “Perching failure detection and recovery with onboard sensing,” in 2015 IEEE / RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2015, pp. 1264-1270.

[0082]

[18] J. Thomas, M. Pope, G. Loianno, E. W. Hawkes, M. A. Estrada, H. Jiang, M. R. Cutkosky, and V. Kumar, “Aggressive fight with quadrotors for perching on inclined surfaces,” Journal of Mechanisms and Robotics, vol. 8, no. 5, p. 051007, 2016.

[0083]

[19] L. Daler, A. Klaptocz, A. Briod, M. Sitti, and D. Floreano, “A perching mechanism for flying robots using a fibre-based adhesive,” in 2013 IEEE International Conference on Robotics and Automation. IEEE, 2013, pp. 4433-4438.

[0084]

[20] A. Kalantari, K. Mahajan, D. Ruffatto, and M. Spenko, “Autonomous perching and take-off on vertical walls for a quadrotor micro air vehicle,” in 2015 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2015, pp. 4669-4674.

[0085]

[21] R. Cory and R. Tedrake, “Experiments in fixed-wing uav perching,” in AIAA Guidance, Navigation and Control Conference and Exhibit, 2008, p. 7256.

[0086]

[22] J. Moore and R. Tedrake, “Control synthesis and verification for a perching uav using lqr-trees,” in 2012 IEEE 51st IEEE Conference on Decision and Control (CDC). IEEE, 2012, pp. 3707-3714.

[0087]

[23] J. F. Roberts, J.-C. Zufferey, and D. Floreano, “Energy management for indoor hovering robots,” in 2008 IEEE / RSJ International Conference on Intelligent Robots and Systems. IEEE, 2008, pp. 1242-1247.

[0088]

[24] M. Graule, P. Chirarattananon, S. Fuller, N. Jafferis, K. Ma, M. Spenko, R. Kornbluh, and R. Wood, “Perching and takeoff of a robotic insect on overhangs using switchable electrostatic adhesion,” Science, vol. 352, no. 6288, pp. 978-982, 2016.

[0089]

[25] P. Sanchez-Cuevas, G. Heredia, and A. Ollero, “Multirotor uas for bridge inspection by contact using the ceiling effect,” in 2017 International Conference on Unmanned Aircraft Systems (ICUAS). IEEE, 2017, pp. 767-774.

[0090]

[26] C. Powers, D. Mellinger, A. Kushleyev, B. Kothmann, and V. Kumar, “Influence of aerodynamics and proximity effects in quadrotor flight,” in Experimental robotics. Springer, 2013, pp. 289-302.

[0091]

[27] V. J. Rossow, “Effect of ground and / or ceiling planes on thrust of rotors in hover,” 1985.

[0092]

[28] P. Chirarattananon and Y. H. Hsiao, “Ceiling effects for hybrid aerial-surface locomotion of small rotorcraft,” IEEE / ASME Transactions on Mechatronics, 2019.

[0093]

[29] C. E. Doyle, J. J. Bird, T. A. Isom, C. J. Johnson, J. C. Kallman, J. A. Simpson, R. J. King, J. J. Abbott, and M. A. Minor, “Avian-inspired passive perching mechanism for robotic rotorcraft,” in 2011 IEEE / RSJ International Conference on Intelligent Robots and Systems. IEEE, 2011, pp. 4975-4980.

[0094]

[30] C. E. Doyle, J. J. Bird, T. A. Isom, J. C. Kallman, D. F. Bareiss, D. J. Dunlop, R. J. King, J. J. Abbott, and M. A. Minor, “An avian-inspired passive mechanism for quadrotor perching,” IEEE / ASME Transactions on Mechatronics, vol. 18, no. 2, pp. 506-517, 2012.

[0095]

[32] W. Chi, K. Low, K. H. Hoon, and J. Tang, “An optimized perching mechanism for autonomous perching with a quadrotor,” in 2014 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2014, pp. 3109-3115.

[0096]

[33] C. Luo, L. Yu, and P. Ren, “A vision-aided approach to perching a bio-inspired unmanned aerial vehicle,” IEEE Transactions on Industrial Electronics, vol. 65, no. 5, pp. 3976-3984, 2017.

[0097]

[34] M. A. Erbil, S. D. Prior, and A. J. Keane, “Design optimisation of a reconfigurable perching element for vertical take-off and landing unmanned aerial vehicles,” International Journal of Micro Air Vehicles, vol. 5, no. 3, pp. 207-228, 2013.

[0098]

[35] Z. Zhang, P. Xie, and O. Ma, “Bio-inspired trajectory generation for uav perching,” in 2013 IEEE / ASME International Conference on Advanced Intelligent Mechatronics. IEEE, 2013, pp. 997-1002.

[0099]

[36] H. Gu, X. Cai, J. Zhou, Z. Li, S. Shen, and F. Zhang, “A coordinate descent method for multidisciplinary design optimization of electric-powered winged UAVs,” in 2018 International Conference on Unmanned Aircraft Systems (ICUAS). IEEE, 2018, pp. 1189-1198.

[0100]

[37] X. Lyu, H. Gu, Y. Wang, Z. Li, S. Shen, and F. Zhang, “Design and implementation of a quadrotor tail-sitter VTOL UAV,” in 2017 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2017, pp. 3924-3930.

[0101]

[38] Y. Wang, X. Lyu, H. Gu, S. Shen, Z. Li, and F. Zhang, “Design, implementation and verification of a quadrotor tail-sitter VTOL UAV,” in 2017 International Conference on Unmanned Aircraft Systems (ICUAS). IEEE, 2017, pp. 462-471.

[0102]

[39] H. Gu, X. Lyu, Z. Li, S. Shen, and F. Zhang, “Development and experimental verification of a hybrid vertical take-off and landing (VTOL) unmanned aerial vehicle (UAV)," in 2017 International Conference on Unmanned Aircraft Systems (ICUAS). IEEE, 2017, pp. 160-169.

[0103]

[40] F. Zhang and J. Song, “Real-time calibration of gyro-magnetometer misalignment,” IEEE Robotics and Automation Letters, vol. 3, no. 2, pp. 849-856, 2018.

[0104]

[41] J. Lin and F. Zhang, “Loam livox: A fast, robust, high-precision lidar odometry and mapping package for lidars of small fov,” arXiv preprint arXiv:1909.06700, 2019.

[0105]

[42] ____, “A fast, complete, point cloud based loop closure for lidar odometry and mapping,” arXiv preprint arXiv:1909.11811, 2019.

[0106]

[43] F. Zhang, X. Lyu, Y. Wang, H. Gu, and Z. Li, “Modeling and flight control simulation of a quadrotor tail sitter VTOL UAV,” in AIAA Modeling and Simulation Technologies Conference, 2017, pp. 1561-1573.

[0107]

[44] X. Lyu, H. Gu, J. Zhou, Z. Li, S. Shen, and F. Zhang, “A hierarchical control approach for a quadrotor tail-sitter VTOL UAV and experimental verification,” in 2017 IEEE / RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017, pp. 5135-5141.

[0108]

[45] J. Zhou, X. Lyu, Z. Li, S. Shen, and F. Zhang, “A unified control method for quadrotor tail-sitter UAVs in all flight modes: Hover, transition, and level flight,” in 2017 IEEE / RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2017, pp. 4835-4841.

[0109]

[46] X. Lyu, J. Zhou, H. Gu, Z. Li, S. Shen, and F. Zhang, “Disturbance observer based hovering control of quadrotor tail-sitter VTOL UAVs using H-infsynthesis,” IEEE Robotics and Automation Letters, vol. 3, no. 4, pp. 2910-2917, 2018.

[0110]

[47] J. Lin, L. Wang, F. Gao, S. Shen, and F. Zhang, “Flying through a narrow gap using neural network: an end-to-end planning and control approach,” arXiv preprint arXiv:1903.09088, 2019.

[0111]

[48] K. Hang, X. Lyu, H. Song, J. A. Stork, A. M. Dollar, D. Kragic, and F. Zhang, “Perching and resting - a paradigm for uav maneuvering with modularized landing gears,” Science Robotics, vol. 4, no. 28, p. eaau6637, 2019.

[0112]

[49] “Px4 sapog,” available at https: / / github.com / PX4 / sapog.

[0113]

[50] M. Cutler and J. How, “Actuator constrained trajectory generation and control for variable-pitch quadrotors,” in AIAA Guidance, Navigation, and Control Conference, 2012, p. 4777.

[0114]

[51] D. Mellinger and V. Kumar, “Minimum snap trajectory generation and control for quadrotors,” in 2011 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2011, pp. 2520-2525.

[0115]

[52] M. W. Mueller, M. Hehn, and R. D’Andrea, “A computationally efficient motion primitive for quadrocopter trajectory generation,” IEEE Transactions on Robotics, vol. 31, no. 6, pp. 1294-1310, 2015.

[0116]

[53] J. Zhou, X. Lyu, X. Cai, Z. Li, S. Shen, and F. Zhang, “Frequency domain model identification and loop-shaping controller design for quadrotor tail-sitter vtol uavs,” in 2018 International Conference on Unmanned Aircraft Systems (ICUAS). IEEE, 2018, pp. 1142-1149.

[0117]

[54] W. Xu, H. Gu, Y. Qin, J. Lin, and F. Zhang, “Full attitude control of an efficient quadrotor tail-sitter vtol uav with flexible modes presentation format: Oral presentation,” in International Conference on Unmanned Aerial Systems (ICUAS), 2019.

[0118]

[55] X. Liang, M. Zheng, and F. Zhang, “A scalable model-based learning algorithm with application to uavs,” IEEE control systems letters, vol. 2, no. 4, pp. 839-844, 2018.

[0119]

[56] G. J. Leishman, Principles of helicopter aerodynamics with CD extra. Cambridge university press, 2006.

[0120]

[57] D. P. Bertsekas, “Nonlinear programming,” Journal of the Operational Research Society, vol. 48, no. 3, pp. 334-334, 1997.

[0121]

[58] J. H. Gillula, H. Huang, M. P. Vitus, and C. J. Tomlin, “Design of guaranteed safe maneuvers using reachable sets: Autonomous quadrotor aerobatics in theory and practice,” in 2010 IEEE International Conference on Robotics and Automation. IEEE, 2010, pp. 1649-1654.

[0122]

[59] S. Lupashin, A. M. Sherback, and R. D’Andrea, “A simple learning strategy for high-speed quadrocopter multi-flips,” in 2010 IEEE international conference on robotics and automation. IEEE, 2010, pp. 1642-1648.

[0123]

[60] F. Oliva-Palomo, A. Sanchez-Orta, P. Castillo, and H. Alazki, “Nonlinear ellipsoid based attitude control for aggressive trajectories in a quadrotor: Closed-loop multi-flips implementation,” Control Engineering Practice, vol. 77, pp. 150-161, 2018.

[0124]

[61] P. E. Tanner, A. D. Overmeyer, L. N. Jenkins, C.-S. Yao, and S. M. Bartram, “Experimental investigation of rotorcraft outwash in ground effect,” 2015.

[0125]

[62] Intel, “Intel realsense d400 series product family.” 337029-005 datasheet, Jan. 2019, rev. 5 (available at https: / / www.intel.com / content / dam / support / us / en / documents / emerging-technologies / intel-realsense-technology / Intel-RealSense-D400-Series-Datasheet.pdf).

[0126]

[63] J.-E. Deschaud and F. Goulette, “A fast and accurate plane detection algorithm for large noisy point clouds using filtered normals and voxel growing,” in 3DPVT, 2010.

[0127]

[64] STMicroelectronics, "World's smallest time-of-flight ranging and gesture detection sensor." DocID029104 datasheet, Apr. 2018, rev. 2 (available at https: / / www.st.com / resource / en / datasheet / vl53l0x.pdf).

Claims

1. A multi-copter unmanned aerial vehicle, comprising: a frame; a plurality of variable speed motors spaced around the frame, each motor driving a propeller to rotate, such that the rotating propellers enable the frame to assume a horizontal hover position above the ground in an environment; a sensor to detect distance between the UAV and a target surface in the environment; a computer executing software modules to control operation of the motors to cause the UAV to perform flight maneuvers; and a computer memory to store the software modules, wherein the stored modules, when executed by the computer, cause the UAV to at least (a) approach the target surface, (b) utilize a "ceiling effect" to contact the target surface in order to perch on the surface, thereby increasing hover accuracy of the UAV and reducing energy consumption of the UAV, and (c) disengage from the target surface, wherein the target surface is one of a ceiling, a substantially vertical wall, the ground, and a slope in the environment, wherein, when in the hover position, the frame has a substantially horizontal planar configuration with a lower side facing the ground and an upper side facing the ceiling, wherein, to perch on a substantially vertical wall or a slope, the UAV (a) ascends a portion toward the wall by increasing speed of the motors above hover speed, (b) flips the UAV sideways by rotating the motors on one side of the frame faster than the other, (c) detects proximity to the wall with the sensor, (d) decreases speed of the motors as the wall is approached, (e) reverses the direction of rotation before contact with the wall to slow contact with the wall, (f) contacts the wall by reversing the motors via the ceiling effect, and (g) disengages from the wall by reversing the direction of the motors, or to perch on the ground, the UAV (a) ascends a portion away from the ground by increasing speed of the motors above hover speed, (b) flips the UAV completely by rotating the motors on one side of the frame in one direction and the other side of the frame in another direction, (c) reverses both sides of the motors, (d) detects proximity to the ground with the sensor, (d) decreases speed of the motors as the ground is approached, (e) reverses the direction of rotation of the motors before contact with the ground to contact the ground via the ceiling effect, and (g) disengages from the ground by reversing the direction of the motors. to perch on a ceiling, the UAV (a) ascends toward the ceiling by increasing speed of the motors above hover speed, (b) detects proximity to the ceiling with the sensor, (c) decreases speed of the motors as the ceiling is approached, (d) decreases rotation of the motors even more as contact with the ceiling is made, but remains in contact with the ceiling due to the ceiling effect, and (e) disengages from the ceiling by reversing the direction of rotation of the motors.

2. The multicopter unmanned flying vehicle of claim 1, wherein, ​ 3. The multicopter unmanned flying vehicle of claim 1, further comprising: A passive omni-wheel located on the ground-facing side and wherein the omni-wheel provides force support from the ground allowing the propeller to spin at much lower speeds and consume less energy.

4. The multicopter unmanned flying vehicle of claim 1, further comprising: A propeller protector located around the propeller wherein contact between the UAV and the target surface is achieved through the propeller protector as a landing structure.

5. The multicopter unmanned flying vehicle of claim 4, further comprising: A damper provided on the protector wherein contact between the UAV and the target surface is achieved through the damper, the damper absorbs residual kinetic energy when hitting a surface, the damper increases friction with the surface during parking, and the damper improves compliance when contacting uneven surfaces.

Images