The disclosure relates generally to robots and more particularly to a jumping robot.
Many robots with jumping ability are known. For these traditional designs, jumping is accomplished by an instant release of the energy stored in the robot. As a result, all the robots with jumping ability can be classified by their energy storage methods. The most popular conventional method to store the energy is based on traditional springs such as compression, extension, or torsion springs. One exemplary jumping robot uses a motor-driven ball screw to charge and release a compression spring. In another example, the energy in an extension spring is stored and released through a geared six-bar mechanism. Yet a further example utilizes a slip gear system to store and release the energy in an extension spring via a four-bar mechanism.
Still another conventional robot employs a planetary gear train and a one-way clutch, where a conical spring directly strikes the ground. Prior prototype robots employ a motor driven eccentric cam to charge a torsion or two extension springs that actuates rear legs. A wheel-based, stair-climbing robot with soft landing ability has a jumping mechanism based on four compression springs. Another exemplary version, known as the EPFL jumper, can achieve a jumping height about 1.4 m with torsion springs charged and released by a motor driven cam system; see Kovac, M. et al., “A Miniature 7 g Jumping Robot,” Proc. IEEE Int. Conf. Robot. Autom. (2008) at 373-378. This robot was later modified to add self-recovery capability and jumping direction changing ability; see Kovac, M. et al., “Steerable Miniature Jumping Robot,” Auton. Robots, Vol. 28, No. 3 (2010) at 295-306; and Kovac, M. et al., “A Miniature Jumping Robot with Self-Recovery Capabilities,” Proc. IEEE/RSJ Int. Conf. Intell. Robots Systems (2009) at 583-588. In another construction, a multimodal robot can jump up to 1.7 m based on two symmetrical extension spring-actuated four-bar mechanisms.
A second traditional method for energy storage is based on elastic elements such as customized special springs. A scout robot employs a motor-driven winch to charge a single bending plate spring and release it to directly strike the ground for jumping. A conventional compact jumping robot utilizes an elastic strip to form closed elastica, which is actuated by two revolute joints. With two symmetrical power springs made of carbon fiber strips as legs, another conventional microbot charges its springs with dielectric elastomer actuators. In another example, known as the Jollbot, a spherical structure formed by several metal semi-circular hoops is used to store energy in the hoops by deforming its spherical shape. A similar conventional idea is utilized in a deformable robot, but the material for hoops is replaced by shape memory alloy.
A third conventional method to store energy for jumping is based on compressed air. In this method, a robot usually carries an air tank and a pneumatic cylinder. The sudden release of air in the tank will force the cylinder to extend. A traditional rescue robot and a patrol robot employ cylinder extension to strike the ground for jumping. Instead of striking the ground, the jumping ability of another traditional robot, known as a quadruped Airhopper, is accomplished by several cylinder-actuated four-bar mechanisms. With a biped structure, another exemplary robot, different from other pneumatic-based jumping robots, uses several pneumatic artificial muscles for jumping.
In addition to the preceding, there exist several other known devices. A pendulum jumping robot is based on the principle that momentum will be generated from swinging arms during human jumping. Furthermore, a jumping robot developed by the Sandia National Labs uses the energy from hydrocarbon fuels and can achieve a significant jumping height. Another robot is based on microelectromechanical (“MEMS”) technology and has a small jumping height. Finally, an additional robot uses a voice coil actuator to charge energy into an electrical capacitor instead of a mechanical structure.
It is noteworthy, however, that traditional jumping robots are either too heavy, too expensive or too large. Furthermore, most conventional robots disadvantageously require multiple actuators which quickly use considerable battery power. While such robots may be needed for space exploration, their weight and size prevent them from achieving desired jumping heights and distances.
In accordance with the present invention, a jumping robot is provided. In another aspect, a jumping robot weighs less than 50 grams, jumps at least 20 cm high and has a maximum linear dimension of 10 cm. A further aspect provides a robot that employs a single electromagnetic actuator that actuates at least two of: jumping, steering and/or self-righting. Still another aspect employs multiple jumping robots that communicate with a remote communications station and/or each other either airborne or on the ground.
The present jumping robot is advantageous over traditional devices. For example, the present jumping robot is light weight and of small size thereby allowing it to achieve long jumping distances or large jumping heights. Furthermore, the present jumping robot is multifunctional in its movements driven by a single actuator; this assists with realizing the light weight, small size and low cost while requiring minor battery power such that more than 100 jumps can be made on a single battery charge. The present jumping robot is ideally suited for mass production so many of them can be used at a natural disaster site or in a war zone to locate victims or targets, respectively. The low cost nature avoids the need for recovery of the robot after use while allowing for many robots to be employed at a single site. Moreover, the significant jumping height allows for line-of-sight and/or airborne sensing and communications over rubble and other ground obstacles. The height of the jump provides a greater sensing and communications coverage area as well. Additional advantages and features of the present invention will become apparent from the following description and appended claims, taken in conjunction with the accompanying drawings.
The preferred embodiment of a jumping robot 31 as shown in
A. Jumping Mechanism:
Reference should be made to
The jumping mechanism, shown in
|AB|=l1,|BC|=|AF|=l2,|CD|=|FE|=l3,and |DE|=l4. (1)
A vertical distance y is between AB and ED, and the angle α is between {right arrow over (AB)} and {right arrow over (BC)}, and an angle β is between {right arrow over (DC)} and X. The mechanism has three planar degrees of freedom from the Kutzbach-Grübler mobility formula. Nevertheless, if eight torsion springs with spring constant k are placed at A, B, E, and D, two springs for each place, then the springs can be charged to store energy if a vertical downward force F is applied at point O′, and the energy is released once F is removed.
For energy storage, an analysis of how much force F, varying with y, is needed to charge the spring is as follows. Since the mechanism is symmetric with respect to OO′, a static analysis for one side is sufficient. Free body diagrams for links AB, BC, and CD are illustrated in
where τ1 and τ2 are the torques generated by the springs. From above equations, F can be solved as:
Note that α and β are functions of y and point C's vertical coordinates yc. Point C is the intersection point of two circles with centers at B: (l1/2, y) and D: (l4/2, 0); therefore, yc can be solved as:
where e=ld2/4+y2 and ld=l4−l1. In fact, there are two intersection points for those two circles, but the point corresponding to the configuration shown in
Substituting them into equation (4), F is expressed as a function of y. To facilitate optimization ymax and ymin are the maximum and minimum value of y. The largest value for ymax is √{square root over ((l2+l3)2−ld2/4)}
when AF and FE, BC and CD are collinear. Although ymax=√{square root over ((l2+l3)2−ld2/4)}, this case is singular, which should be avoided. To simplify the design process, let
ymax=0.95√{square root over ((l2+l3)2−ld2/4)}. (7)
B. Energy Mechanism:
Based on the jumping mechanism, another energy mechanism is needed to store energy in the spring and release it when necessary. Referring to FIGS. 1 and 8-10, a single direction rotation of actuator 41, which is an electric motor, is used for energy storage and release, leading to a short cycle time. This can be achieved by a slip-gear system, an eccentric cam or a variable length crank mechanism. Preferably, a one way bearing is employed. The energy mechanism, shown in an intermediate position in
With such a mechanism, a vertical downward force is applied for energy storage. For optimization this force is related to torque generated by a speed reduction system. This can be achieved by static analysis for crank 91. As can be observed in
Note that F is expressed as a function of y in the static analysis for the jumping mechanism. But y varies with φ and the relation between them is:
y=ymax−(√{square root over (ld2+(la+lb)2−2la(la+lb)cos φ)}{square root over (ld2+(la+lb)2−2la(la+lb)cos φ)}−lb) (9)
C. Self-Righting Mechanism:
With the above two mechanisms, the robot can jump if the robot initially stands with its foot on the ground. Nevertheless, after the robot jumps into the air and lands on the ground, this optimal orientation may not occur due to the landing impact. Therefore, as illustrated in
The details of a driving device 141 are shown in
Referring now to
D. Steering Mechanism:
The final mechanism to realize the jumping motion sequence is steering mechanism 37, which can change the jumping direction without requiring an additional actuator. The steering mechanism is based on the speed reduction transmission in the energy mechanism. Since all the above-mentioned motions can be achieved using one directional rotation, the steering mechanism can be actuated by the same multi-functional motor actuator 41, but with rotation in another direction.
Alternately, a single large gear at the end of the speed reduction transmission acts as the steering gear. Due to its large diameter, the alternate steering gear touches the ground beyond the outer robot surfaces regardless of which largest surface of the robot contacts the ground. This method, although simpler, has a slow steering speed because the angular velocity of the large gear is small. The preferred design, however, solves this disadvantage by using two gears next to the motor output pinion gear so that their angular velocities are larger.
E. Optimal Design of Jumping Mechanism and Energy Mechanism:
With the jumping mechanism and the energy mechanism, if the same energy in the spring can be charged with a small peak value of torque T generated by the speed reduction system, then the weight and size for the speed reduction system can be reduced. Therefore, mechanism optimization is desirable to minimize the peak value of T. The optimal design calculations are set forth as follows. The optimization is achieved following the procedure of identifying the optimization variable, formulating the objective function, obtaining the constraints, and solving the constrained optimization problem. To perform an optimal design, the optimization variables should be first identified. Substituting force equation (4) into the torque equation (8):
from which there are eight parameters k, la, lb, α, β, l2, l3, and φ. Since α and β are related to y and yc through equation (6) and yc is in turn related to y, l2, l3, and ld via equation (5), the true parameters are k, la, lb, y, l2, l3, ld and φ.
Among the above eight parameters, the variables will be only lb, l2, l3, ld, and φ. In other words, k, la, and y are either constants or can be obtained from l1, l2, l3, ld, and φ. Furthermore, the torsion springs are chosen to have a constant k=58.98 Nmm/rad (preferably obtained from McMaster Carr as part no. 9287K77). Moreover, la is represented as a function of l2, l3, and ld as will be shown in what follows. Let ymin be the final minimum distance for y when the spring is fully charged, then la=(ymax−ymin)/2. Since ymax is related to l2, l3, and ld as shown in equation (7), ymin can also be expressed as a function of them if the initial energy E0 is known. Based on the simulations, to achieve one meter jumping height with a 75° take-off angle, the initial energy should be 0.3 J. But to leave some margin, let the initial energy be E0=0.4 J. Then ymin can be obtained from:
where α1 and β2 are the minimum angles for α and β when y=ymin in equations (5) and (6), and α2 and β2 are the maximum angles when y=ymax. Note that an analytical expression for ymin is unavailable because of the trigonometric items; however, a numerical solution can be obtained once l2, l3, and ld are given. Therefore, la can be obtained from l2, l3, and ld. Finally, y can also be derived from l2, l3, and ld through equation (9). Based on the preceding, T is only a function of lb, l2, l3, ld, and φ it is denoted as T(lb, l2, l3, ld, φ). The optimization variables are only lb, l2, l3, and ld because φ will run from 0 to π during each energy charge cycle.
Having identified the optimization variables, the next step is to formulate an objective function. Given lb, l2, l3, and ld a torque curve as φ running from 0 to π can be plotted. The goal is to find the optimal lb, l2, l3, and ld such that the resulting curve has the property that the largest torque in the curve is minimum among all possible curves. In this case, the objective function is the peak torque in the curve for given lb, l2, l3, and ld, which can be defined by:
The next step is to obtain the constraints for the optimization variables. The lengths of l2 and l3 should not be too small, otherwise, it cannot hold the torsion springs. At the same time, they cannot be too large because of the size limit of the robot. Therefore, with practical consideration, assume 15 mm≦l2, l3≦20 mm. With similar implementation reasons, other linear constraints for lb and ld, and the optimization can be formulated as:
minimize g(lb,ld,l2,l3)
subject to 7≦lb≦12,−5≦ld≦5,15≦l2≦20,15≦l3≦20 (13)
where the omitted length unit is millimeters. To solve the constraint optimization problem, the numerical method is implemented in Matlab because the analytical expression for g(lb, la, l2, l3) cannot be obtained. The optimization is performed by dense discretization of φ and evaluating the value at the resulting points. The constrained nonlinear multivariable function in the Optimization Toolbox is employed to find the optimal value. Since the method can only obtain the local minimum, various random initial points are chosen to run the optimization. The optimization results are the same for these initial points, and the optimal values for the mechanisms are lb=7 mm, ld=1.2 mm, l2=15 mm and l3=20 mm. The other parameters can be calculated accordingly: ymax=33.3 mm, ymin=11.7 mm, and la=10.8 mm. To avoid interference between the two revolute joints at the foot, let l1=18 mm, then l4=l1+ld=19.2 mm.
F. Optimal Design of the Self-Right Mechanism:
The self-righting mechanism is designed to successfully recover from any landing posture. The design variables include the arm length and the range of arm rotation angle. The initial and final positions for both recovery arms 131 are shown in
From above equations, if μ or v is large, then the arm length AA′ or BB′ will be small. To simplify the design, μ=135° and v=105° are fixed. Such a choice is made to let the difference between μ and ∠ODA′, ν and ∠ODB′ be 30°. Then |AA′| and |BB′| be solved from about two equations based on |AB| and |OD|. Note that both |AB| and |OD| are determined from real implementation.
The next step is to design the length of AC shown in
|BC1|=√{square root over (|AB|2+|AC|2−2|AB∥AC|cos(μ/2))}. (16)
Turning to
A three-stage gear train connects an output shaft of motor actuator 41 to crank 91. The gear train is made up of output pinion gear 173 with 8 teeth, compound steering gear 171 with 28/8 teeth, another compound gear 251 with 28/9 teeth, and spur gear 95 with 35 teeth. Therefore, the preferred total speed reduction ratio is 47.6. At both ends of each shaft, a set of miniature ball bearings are used to reduce the friction of the associated gears. Moreover, guidance pulleys 101 are also a set of miniature ball bearings. The robot is preferably powered by a 50 mAh FullRiver LiPo battery 53 (see
Comparing
G. Tail Assisted Orientation Control:
Referring to
A schematic diagram of robot 31 in the air is shown in
Using a Lagrangian formulation, a dynamics equation is obtained for jumping robot 31, illustrated in
M{umlaut over (θ)}t−L cos θm{umlaut over (θ)}b+L sin θm{dot over (θ)}b2=τ (17)
N{umlaut over (θ)}b−L cos θm{umlaut over (θ)}t−L sin θm{dot over (θ)}t2=−τ (18)
where
and τ is the torque generated by motor 311. Since there is only one control input, both θb and θt are not simultaneously controlled. Thus, to eliminate one variable from the equation, the total angular momentum for the system is used which is:
H0=(M−L cos θm){dot over (θ)}t+(N−L cos θm){dot over (θ)}b (20).
With the previous two equations, the dynamics equation for θm is obtained as follows:
With the dynamics equation, a controller is designed for the system to automatically generate the appropriate τ so that the angle θb is stabilized at a desired angle.
A dynamics equation is also used to analyze the required time to reach a desired angle. In this case, the motor is considered to be under a constant voltage supply. Then the dynamics equation can be rewritten as:
where τs is the stall torque and ωn is the no-load angular velocity for motor 311 under the voltage supply. Solving the equation, the trajectory is estimated for θm. For example, with body 33 being about 25 grams, if tail weight 305 has a point mass of three grams at the end of a six centimeters long rod 303, the relative angle θm increases 90 degrees in 0.1 second.
To control the robot with the tail option, an embedded and automated control system is designed with the schematic of
Sensing unit 403 includes a three axis accelerometer 411, a three axis gyroscope 413, and a three axis magnetic compass 415. A single chip is employed for the preceding two sensors (for example, Model No. MPU-6050 from InvenSense, Inc.), while compass 415 is another chip (such as Model No. HMC5883L from Honeywell Inc.). With the accelerometer and gyroscope, a closed loop and automated, real-time feedback control for mid-air maneuvering can be achieved. Specifically, accelerometer 411 detects the free fall, while gyroscope 413 provides feedback of the body's angle to controller 401. Compass 415 is employed to determine the robot's heading direction when the robot lands on the ground.
Actuation unit 405 includes one or more motor drivers with pulse width modulation (“PWM”) ability. In the preferred version, a dual H-Bridge DC motor driver (Model No. MC34933 from Freescale Semiconductor, Inc.) controls both jump motor 41 and tail motor 311. With the PWM ability, commands from the microcontroller control the voltage on the motors, thereby dictating the motor speed.
Robot 31 is powered by a small battery 421 (also shown as 53 in
The tail can only rotate about one axis to control the robot's pitch angle in the illustrated design; however, an alternate design is also envisioned to control the robot body's orientation; this is represented by roll, pitch, and yaw angles. In this alternate case, two motors instead of one need to be employed for the tail function so that the tail can rotate about two axes. It should be appreciated that the present tail assembly may be optionally used in combination with the jumping, steering and self-righting features of the robot shown in
The remote communications device further sends a movement signal to each robot 31 to cause the on-board microprocessor to energize the actuator motor to cause jumping, steering and/or self-righting movements. The on-board microprocessor controller and/or the remote communications device are connected to non-transient memory, such as ROM, RAM, a disk or a removeable memory chip, within which are programmed software instructions 265. The software instructions, as depicted in
While various features of the present jumping robot have been disclosed, other variations can be employed. For example, while two outwardly bending jumping legs have been disclosed, three, four or more bending legs can alternately be used, although some of the light weight and, thus, jumping height advantages may not be achieved. Furthermore, cable 99 and pullies 101 may be replaced by one or more chains or rigid articulated links, although certain advantages may not be realized. Additionally, alternate gearing combinations, steering wheels (instead of gears 171), biasing members, and leg and arm shapes may be utilized, but certain advantages may not be found. Moreover, it is alternately envisioned that the robot can be peripherally coated with a hydrophobic polymer, and provided with larger ground-contacting surface areas (e.g., at the foot), allowing for jumping from a surface of water upon which it rests. Accordingly, such variations are not to be regarded as a departure from the disclosure, and all such modifications are intended to be included within the scope of the present invention.
This application claims the benefit of U.S. Provisional Application Ser. No. 61/625,881, filed on Apr. 18, 2012, which is incorporated by reference herein.
This invention was made with government support under CNS0721441 awarded by the National Science Foundation. The government has certain rights in this invention.
Number | Name | Date | Kind |
---|---|---|---|
1573527 | Simonek | Feb 1926 | A |
2938300 | Newbert et al. | May 1960 | A |
3086738 | Lubash | Apr 1963 | A |
4204359 | Plo | May 1980 | A |
4624648 | Waters | Nov 1986 | A |
4702720 | Konta et al. | Oct 1987 | A |
4894042 | Kamikawa | Jan 1990 | A |
5040626 | Paynter | Aug 1991 | A |
5188315 | Foitzik et al. | Feb 1993 | A |
5413514 | Milligan | May 1995 | A |
5618219 | Simone et al. | Apr 1997 | A |
5964639 | Maxim | Oct 1999 | A |
6247546 | Spletzer et al. | Jun 2001 | B1 |
6484068 | Yamamoto et al. | Nov 2002 | B1 |
7249640 | Horchler et al. | Jul 2007 | B2 |
7270589 | Brown, Jr. et al. | Sep 2007 | B1 |
7434638 | Tanielian | Oct 2008 | B2 |
7481691 | Sze et al. | Jan 2009 | B2 |
7503410 | Delson | Mar 2009 | B2 |
7963351 | Troy | Jun 2011 | B2 |
Number | Date | Country |
---|---|---|
101327817 | Dec 2008 | CN |
201240437 | May 2009 | CN |
201703452 | Jan 2011 | CN |
201769932 | Mar 2011 | CN |
102050156 | May 2011 | CN |
102050157 | May 2011 | CN |
201815109 | May 2011 | CN |
20110000852 | Jan 2011 | KR |
Entry |
---|
Zhao, Jianguo, et al.; “Development of a Controllable and Continuous Jumping Robot,” 2011 IEEE International Conference on Robotics and Automation, May 19-13, 2011, pp. 4614-1419. |
Ackerman, Evan; “Brillant Little Jumping Robot Only Needs One Motor,” ...ieee.org/.../brillant-little-jumping-robot-only-needs-one-motor, posted Jun. 6, 2011, pp. 1-2. |
Ackerman, Evan; “World's Cleverest Jumping Robot Gets Faster, More Agile,” ...ieee.org/.../worlds-cleverest-jumping-robot-gets-faster-mor . . . , posted Nov. 28, 2011. |
Kaur, Kanwalpreet: “A Study of Bio-Inspired Jumping Robots,” International Journal of Computer Applications, vol. 48, No. 12, Jun. 2012, pp. 40-43. |
Zhao, Jianguo et al., “A Single Motor Actuated Miniature Steerable Jumping Robot,” 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, Oct. 7-12, 2012, pp. 4274-4275. |
Kovac, Mirko et al.; “A miniature 7g jumping robot,” http://infoscience.epfl.ch/record/111783?ln=en, Proceedings of IEEE International Conference on Robotics and Automation (ICRA '2008), publicly presented and/or published May 19-23, 2008; pp. 373-378 (7 pages, includng Abstract page). |
Kovac, Mirko et al.; “Steerable miniature jumping robot,” Auton Robot 2010, 28, published Dec. 30, 2009; pp. 295-306. |
Kovac, Mirko, et al.; “A Miniature Jumping Robot with Self-Recovery Capabilities,” http://infoscience.epfl.ch/record/139011?ln=en, Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, 2009, publicly presented and/or published Oct. 11-15, 2009; pp. 583-588 (7 pages, including Abstract page). |
“Draganfly Innovations' New UAV Helicopter:The Draganflyer X6—Industrial 6-Rotor Unmanned Helicopter Provides High Definition Wireless Vido and High Resolution Photographs;” Draganfly Innovations, Inc., published Aug. 2008; 2 pages. |
“Draganfly X6 UAV Helicopter Wins Popular Science Best of What's New Award in Aviation & Space;” Draganfly Innovations, Inc., published Nov. 2008; 2 pages. |
“Sling-Copter—The Amazing Spinning Flying Helicopter Toy;” internet advertisement, Sling Copter, www.slingcopter.com <http://www.slingcopter.com>, published 2011; 3 pages. |
Number | Date | Country | |
---|---|---|---|
20130282174 A1 | Oct 2013 | US |
Number | Date | Country | |
---|---|---|---|
61625881 | Apr 2012 | US |