Real-World Applications

cancel
Showing results for 
Search instead for 
Did you mean: 

The Dynamic Model of the Human-Like Aerial Manipulator

Company: Kalashnikov Izhevsk State Technical University
Author(s): Ibrahim Ibrahim Nizar
NI Product(s) Used: LabVIEW, Control Design and Simulation, Robotics Module
Industry: Robotics, Research


The Challenge

Addressing the dynamic model of the 11-DoF robotic arm prototype which is intended to be installed on the drone for executing remote actions. comparing fundamentally to the popular aerial manipulators, this 2nd generation could do actions like human's actions, that would be more close to people: clients or users, easy to understand the intent of actions, predict next ones, teaching new motions. The goals: Rescue and emergency, help remotely, chef applications, medical services, disabled people, delivery and so on.


The Solution

Determining the dynamic of robots is important to solve the subsequent issues in control and application. but it is difficult somehow when the robot is tree-like and has a high degree-of-freedom. The proposed aerial manipulator was considered as a complex engineering model because it consists of 28 degrees of freedom and described as a human-like arm. therefore, it was divided into two parts as follows: arm part and hand part, this within two principal frames, 28 joints and 23 links as shown in the figure below.

Aerial Manipulator 1st prototypeAerial Manipulator 1st prototype

Now, in order to get the parameters of the kinematic model accurately, I used the item <Acquire Kinematics Parameters VI> in Palette: Serial Arm Definition VIs in the robotics library in LabVIEW. This tool depends on the XML file generated from Solid-works and outputs a D-H table. moreover, by setting the dynamic parameters (see figure below) the manipulator object will be ready to the control part. this figure presents the arm part of manipulator and the hand-part is also done the same way. so we will get six tables of D-H.

Figure. The illustration of dynamic setting of robotic arm in LabVIEW.Figure. The illustration of dynamic setting of robotic arm in LabVIEW.After building the dynamic, the current step will be the verification and validation process. it consists of checking the pose and joints trajectory of the manipulator by a given joints data. it is basically carried out by the forward dynamic with zero torque for all joints then change gradually all the torques of joints and write the results, this process should be done along with the forward kinematics to write the pose data. the figure below shows one page of this process. forward dynamics computes the joint angles and movements of end-effector that result from applying a given amount of torque to all joints over a period of time. the next figure shows the forward dynamics and as presented it depends on the torque function (VI strictly typed reference). Figure. The forward kinematics process of the manipulator using the model robotFigure. The forward kinematics process of the manipulator using the model robot
Figure. The forward dynamic of robotic arm with zero torques..Figure. The forward dynamic of robotic arm with zero torques..

The Mathematical Solution

The dynamic equations were introduced for computing the required torque vector of torque space . these calculations depend on position, velocity, and acceleration of system's angles of joints  by taking into our consideration external forces and moments on the links. for simplicity, the whole dynamic system was divided into two parts according to the workspace. arm's serial motion regarded to basic shoulder frame, and fingers' parallel motion regards to basic hand frames. the iterative dynamic Newton-Euler formula was applied to model the whole system by taking into our attention that each link is a rigid body has a designated center of gravity, mass, and inertia called mass distributing. the friction's factors were involved as Coulomb and Viscous. there are the classical formulas which are iterative outward and inward as follow:NE.PNG

Here, because of the manipulator is aerial, it is subject to disturbances of effects as aerodynamic as wind and so forth, so some formulas were developed which took into consideration added external forces and moments on the links as follows in figure below, these formulas didn't yet implement in LabVIEW robotics. now the research is going to prove it in the next studies.

developed NE.PNG
By supposing i an index of a link as a rigid body with mass m, w and w' are the angular velocity and acceleration of a link respectively, whereas v' is the linear acceleration of a link while v'c is the linear acceleration of the center of mass of the link, F and N are force and torque of the mass center of a link i, f and n are static force and moment exerted on link i by link i-1, P is the link length between links i and i-1 while Pc is location of mass center of link between links i and i-1, finally Ic is the inertia tensor matrix of link i. Now, the developed state space of the entire dynamic manipulator is designated as follow: 
DSS.PNG

Where τ is the n×1 vector of joint torques furnished by the actual actuators, and θ is the n×1 vector of joint positions, the matrix M(θ), is a n×n, symmetric, positive definite matrix called manipulator mass matrix. The vector V(θ,θ') signifies torques resulted from centrifugal and Coriolis forces. Whereas the vector G(θ) denotes the torques arises from the gravity forces, F(θ,θ') expresses the torques resulted from frictions forces and finally D(θ) represents the torques of disturbances induced on each link. this mathematical relation furnishes a clear perception about the variants affected the motion of manipulator as gravity, centrifugal, Coriolis, mass, and friction. moreover, it decreases the complexity when studying the motion with ignoring the angular acceleration as suggested previously in order to conclude a control mechanism with the Jacobean relation. the mechanical system is affected by frictions which are around to 25% of the required torque for each joint caused by a significant gearing mechanism. the following formulas exhibit two types of frictions such as Coulomb and viscous according to position and velocity of each angle.

F.PNGWherever Tvf=vθ' is the model of viscous friction that is proportional to the velocity of joint motion where v is a viscous-friction constant. while Tcf=c sgn(θ') is the model of Coulomb friction which is constant except for an sgn dependence on the joint velocity where c is the Coulomb-friction constant, furthermore Cf is the constant indicates the friction of the position of the joint motion. the diagram in the following figure addresses a good detail about the full mathematical dynamic system and their relationships in addition to inputs.
DM.PNG

The inverse dynamic equation was presented in LabVIEW as a function:

Figure. The programmed ID in LabVIEW robotics.Figure. The programmed ID in LabVIEW robotics.

As shown in figure above, the dynamics here depends on gravity and external forces affected on the last link. we can use other VI in the Palette: Dynamics VIs which depends on inertia or Coriolis forces but it still doesn't take into account the external forces on the links. next we will derive mathematical equations of manipulator with external forces as a disturbances. these forces is acting as wind, aerodynamics or other actions could affect the motion the links in the air.

 

The Solution

The following sections proffer a dynamic study of arm-part. The whole study begins with the basic arm's frame to determine the required torques in order to move the basic hand's frame to a specified location in terms of joint motion space. the next figure illustrates the external forces and moments which reflect the real disturbances behaving on the links. The dynamic of hand-part consists of five dynamical investigations which are analogs but the only difference is the physical parameters. the hand-part wasn't explained in this study. The following figure illustrates the forces and moments acting on the links for the arm-part in order to sketch the dynamical models.

arm.PNGBased on our method with developed dynamic Newton-Euler formulas by taken into our consideration the external forces and moments we can conclude the dynamic model as follows:

armm.PNG

The dynamic of hand-part consists of five dynamical investigations which are analogs but the only difference is the physical parameters. it is apparent that the thumb has a different structure among others so it’s model would be more reduced. the following figure illustrates the forces and moments acting on the links for the fingers in order to sketch the dynamical models.

Figure. Illustration of external forces and moments of fingers where i=9 for the thumb finger, i=29 for the little finger, i=24 for the ring finger, i=19 for the middle finger and i=14 for the index finger.Figure. Illustration of external forces and moments of fingers where i=9 for the thumb finger, i=29 for the little finger, i=24 for the ring finger, i=19 for the middle finger and i=14 for the index finger.

Mathematical equations of the thumb fingerMathematical equations of the thumb finger

Mathematical equations of the index, middle, ring and little fingers.Mathematical equations of the index, middle, ring and little fingers.
Mathematical equations of the hand-partMathematical equations of the hand-part
 

Conclusion

In accordance with the dynamics and its complexity, The LabVIEW program and especially Robotics Module were considered as an ideal tool regarding the easy way in understanding the dynamics, solving it and finally developing it. the amazing in this subject at all is the ability to do the experiments easily by sending the joints data to motors directly using RIO devices through control system on FPGA chip and drive them in a parallel way.
The next case study will discuss the reduced model of the manipulator and its dynamics.

 

This is the first launch of the prototype.

 

The first prototype of this manipulator after enhancing the hand.

 

Additional Materials

https://mech.novtex.ru/jour/article/view/535

https://elibrary.ru/item.asp?id=29162302

http://izdat.istu.ru/index.php/vestnik/article/view/4044

https://www.youtube.com/watch?v=Is0R3t1ZI9I

http://robotrends.ru/pub/1916/skolkovo-robotics-2019-manipulyator-dlya-drona

 

Author Contact Details

Ibrahim Nizar Ibrahim

ibrncfe@gmail.com

Research Engineer

https://www.linkedin.com/in/ibrahim-

Kalashnikov Izhevsk State Technical University
Pesochnaya, 38A, Izhevsk, ISTU, Department of Electronic and Electrical Engineering
Russian Federation

Comments
RER
NI Employee (retired)
on

Cool project. Opens up a lot of exciting new use cases for drones. Thank you for sharing Ibrahim!

Rich Roberts
Senior Marketing Engineer, National Instruments
Connect on LinkedIn: https://www.linkedin.com/in/richard-roberts-4176a27b/
FireGiant
Member
Member
on

Very cool project. I am rusty on my dynamics I'm going to have to read through this a couple times and see what you did. Way to go on calculating all the forward dynamics. I am fairly new to National Instruments but graduated in mechnical so I will have to play with this robotics library. I didn't even know it existed.

Contributors