Motion control Algorithms Essay Example



The software control system facilitates a service robot to interact with the humans, learn new skills and tasks from humans, safely steer in the environment, see obstacles and trail them, make use of robotic armrest, and carry out other roles. The technology on learning of robots is one of the major breakthroughs that have led to development of universal purpose software robotic brain.

The packages of the software are built to function on computers’ control of autonomous multi tasks of robots’ service. The software facilitates building of robots that are autonomous and can multitask. The same robots can also have the ability to learn skills and procedures from the human as they maintain as they get in touch with one another.

The possible algorithms

Basing on the Kim et al proposal on an algorithm for robots in environments of nature 3D. The robot must therefore be of strong suction pads. As it can be observed from the Chestnutt et al idea it is appropriate to use apply a plan that is hierarchical to handle avoidance of obstacles and not the complicated task of scaling over obstacles. Chestnut applied the path plan of 2D and avoidance of obstacles 3D.

In this article, our technique is going to major on the hierarchical reinforcement learning applying two level hierarchical decomposition of the errand. For instance an obstacle such as a step where the robot is expected to scale, the higher level planner decides on the sequence of the target positions to move to. It is therefore the responsibility of the low level controller task to shift the robot to this target in an organized manner, whereas maintaining the balance of the robot and averting it from hitting any obstruction.

The operation of the low level controller is on a short chronological scale, in a case where balanced and excellent coordination is needed.

In such incidences the policy search algorithm with simple parameterized policies finds its use well in problems of this nature; it is applied in learning the low level controller.

On the other hand, the high level controller plans the succession of placement positions and must reason on considerably longer timescales. We have to construct a high level controller applying the search of beams and approximation of value function. This is where, the desirability of the various states in the problem are indicated by the value of function.

It is worth noting that we get to discern the value function through the use of novel reinforcement learning algorithm, this also makes use of information that is supervised. The beam search beam is then used in conjunction with learned value function to establish the succession of robot placement that capitalizes on the reward. This succession is then implemented by the low level controller. Hierarchical reinforcement learning algorithms as in the past been used on many problems such as, variants of grid world.

In a nut shell the algorithms that are going to be applied here are Model based control, Low level motion controls, and Learning from demonstration.

The concepts

System controller

  • Task level control

  • Task level control with DLS

  • Joint level control

  • Singular conditions

  • Coriolis, configural terms and gravity

  • Singular terms

Position estimation

  • Real time model based Position estimation

  • Odometry

  • Robot kinematics

Robot controller

  • Reference trajectory generator

  • Position controller


  • Robot dynamics

  • Robot kinematics

Path planner

  • Desired state of the segment

  • End of the segment

Control software

  • Closed loop motion control

  • Low level trajectory generation

  • Reference vector

Robot controller

  • Front steering angle

  • Drive velocity

Classification tree

Front steering angle

Drive velocity

Reference vector

End of the segment

algorithms for robot motion controlalgorithms for robot motion control 1algorithms for robot motion control 2algorithms for robot motion control 3algorithms for robot motion control 4algorithms for robot motion control 5algorithms for robot motion control 6

Path planner

obot controller algorithms for robot motion control 12algorithms for robot motion control 11algorithms for robot motion control 10algorithms for robot motion control 9algorithms for robot motion control 8algorithms for robot motion control 7R

Closed loop motion control

Robot dynamics

Desired state of the segment

Robot kinematics

Robot dynamics

Position controller

ontrol softwarealgorithms for robot motion control 13

Low level trajectory generation



Reference trajectory generator

Robot controller

Robot kinematics


osition estimation algorithms for robot motion control 14P

Real time model based position estimation

Task control, task level control, and joint level control

algorithms for robot motion control 15

ystem controller

Singular conditions, coriolis, configural terms and gravity, and singular terms

algorithms for robot motion control 16S

In the concepts and classification trees the plan of the path is from the off board planner of the path consisting of a record which spells out the path’ segment. The specification of the segment entails the type of the segment at the preferred state of the segments’ end. The specifications of the segment are launched to the control software of the low level trajectory generation and closed loop control of motion.

The robot controller controls the angle of the frontage steering and the velocity of the drive applying the compensation of conventional feed back to retain small slip-ups between the measured states and the reference.

The robot controller receives apposite states of reference from the trajectory generator for it to track. The generator also works out at every control cycle of updates the speed of drive and the steering angle, the gap to the end of the immediate segment of the path, and the state of the reference to the end of the path.

The angle of steering of the front wheel and the speed drive of the front wheel relies on the dynamics of the load motor in reaction to the control signal to the motor. The controller of the path spawns the steering and the signals of commands of the drive.


Jialun, Y, Feng, G, Zhelin, J, & LiFeng, S. (2009). SCIENCE IN CHINA PRESS. Classification of lying states for the humanoid robot SJTU-HR1 , 1-9.

Stilman, M, Michel, P, Chestnutt, J, Nishiwaki, K, Kagami, S, & Kuffner, J. J. (2004). Augmented Reality For Robot Development and Experimentation. Tokyo: Digital Human Research Center, National Institute ofAdvanced Industrial Science and Technology (AIST).

Tan, J, Xi, N, & Wang, Y. (2004). A singularity- free motion control algorithm for robot manipulators- a hybrid approach system . Shenyang: ELSEVIER.