Abstract

In  this  paper,  a  new  Real  Time  Dynamic  Proportional  Integrated  Derivative  approach  “RTDPID” controlled by neural networks for non-holonomic wheeled mobile robot control is  presented.  The  structure  of  the  PID  and  the  parameters  that  intervene  in  the  control  process  are  real  time  adjusted  depending  on  the  trajectory  type  and  obstacles  detected.  The  new  proposed  method  is  compared  with  the  classical  PID  controller  and  has  proved  high efficiency and precision in real time control, especially when evaluated in a hostile environment, which leads to an autonomous decision making to cross over an obstacle or avoid it and stop the engine to protect it. The control process takes into account disturbances appearing during the mobile robot assignments evaluated in a hostile environment, reacts quickly, and adapts to changing environment conditions. In fact, both robot and the control process,  faced  with  obstacles  such  as  holes  or  stones,  are  autonomous  when  it  comes  to  decision making. In case the obstacles are insurmountable, it is necessary to take a decision and stop the engine to avoid damaging it. In addition, an implementation of a kinematic and a dynamic model based on Lagrangien mechanics, are presented regarding non-holonomic and  rolling  without  sliding  constraints.  This  implementation  has  been  developed  and  stimulated in real time on Matlab/Simulink environments. Incidentally, simulation results are given based on the effect of load torque applications on the mobile robot behavior. These simulation results are satisfactory and have demonstrated the effectiveness, robustness and high stability of the proposed technique.

Keywords

robotics, pid controller, hostile environment, neural networks, wheel differential drive, mobile robot, kinematic model, dynamic model, tracking control, speed control, mechatronic system, dynamic, algorithm, simulations, constraints