Balanced Postural Control and Path Planning Analysis of a Two-Wheeled Mobile Robot Using Various Soft Computing Techniques

Chhotray, Animesh (2019) Balanced Postural Control and Path Planning Analysis of a Two-Wheeled Mobile Robot Using Various Soft Computing Techniques. PhD thesis.

[img]PDF (Restricted upto 05/07/2021)
Restricted to Repository staff only

10Mb

Abstract

Two-Wheeled Mobile Robot (TWMR) is a highly nonlinear, multivariable and statically unstable system due to its under actuated configuration. This nonlinear dynamic behaviour of its mechanical system creates a highly coupled condition. Therefore, autonomous collision free navigation of the TWMR while simultaneously maintaining its inverted stable upright posture is really challenging in nature. Here, first of all, the TWMR’s position and orientation in local co-ordinate system with respect to the global co-ordinate system is determined by the kinematic analysis considering its linear and angular velocities. Subsequently, through dynamic modelling, a linearised state space representation of the nonlinear robotic system considering its DC motor dynamics is established by using Newton’s method. From this state space representation, various control parameters like angular tilt, rate of tilt, linear position and velocity of the inverted mobile platform of the TWMR can be determined. After obtaining the raw sensory information from the accelerometer and gyroscope readings, a two stage noise filtration is done by implementing complementary as well as Kalman filter to acquire clear, reliable and accurate data about the internal state of the system. It has been observed that the sensor fusion readings by implementing Kalman filter shows better results than the complementary filter by presenting a better approximation of gyro bias and drift. Two different balanced postural control approaches such as Cascaded PI-PD integrated PID controller and Cascaded Fuzzy-Fuzzy integrated PID controller have been developed for balancing the upright posture of the intermediate body by controlling the requisite torque input to the robot plant model. From the graphical representation of the systemic attitude estimation, it is clearly evident that accurate tilt, linear position and yaw orientation as desired is achieved by the Cascaded Fuzzy-Fuzzy integrated PID controller. After that a novel path planning methodology, DAYANI Arc Contour Intelligent Technique (DACI) has been developed and implemented on the TWMR for determining the best next feasible position in the work space for its continuous motion from the start position to goal position by avoiding collision with intermediate obstacles. The average percentage of error as obtained by comparing the results about the traversed path length and time taken by executing this algorithm in both simulation and experimental modes is found to be about 7%. Further enhancement of the results has been acquired by introducing two meta-heuristic approaches integrated with modified DAYANI method. Here, both Mamdani Fuzzy inference method and back propagation Neural Network method have been deployed for calculating the global scale factor for each probable position from the arc contour of the DAYANI approach. The robot moves subsequently to the position associated with maximum value of the global scale factor and the process continues until the target position is reached. Finally, by the development of a hybrid modified DAYANI-Fuzzy-Neuro-PSO technique, the percentage of error is reduced to below 5% for both path length and travel time taken. Also authenticities of the proposed controllers are verified through a series of comparative analysis with the results obtained from other existing techniques in the similar set of environmental scenarios. Substantial enhancement of results have been observed with the use of developed hybrid controllers in comparison to other results.

Item Type:Thesis (PhD)
Uncontrolled Keywords:Two-Wheeled mobile robot; Dynamic modelling; Sensor fusion; Stability analysis; DACI technique; Modified DAYANI fuzzy; Back-propagation neural network; Hybrid modified DAYANI-fuzzy-neuro-PSO.
Subjects:Engineering and Technology > Mechanical Engineering > Robotics
Engineering and Technology > Mechanical Engineering > Thermodynamics
Engineering and Technology > Mechanical Engineering > Computational Fluid Dynamics
Divisions: Engineering and Technology > Department of Mechanical Engineering
ID Code:10012
Deposited By:IR Staff BPCL
Deposited On:28 Jun 2019 17:03
Last Modified:28 Jun 2019 17:03
Supervisor(s):Parhi, Dayal R.

Repository Staff Only: item control page