ABSTRACT
A robot-assisted control system based on the vector model is proposed for a wheeled mobile robot. According to the closed-loop control structure, the system is constituted of two parts – localization and path planning. The localization algorithm, which is enhanced from Monte Carlo localization, is more effective, stable, and robust than the traditional algorithm because of using many strengthening mechanisms, such as using a vector model, re-initialization, and reverse convergence. The path planning algorithm includes three stages to obtain a path with a motion plan. Firstly, a path from the current position to the goal is planned by an enhanced A* algorithm. Secondly, a smooth mechanism is applied to the path to obtain the continuity of orientation. Finally, a motion design based on the trapezoidal-curve velocity profile is implemented to the smoothed path in both linear and angular velocities to obtain the estimated moving time, position schedule, and velocity schedule. With the assisted control system, the robot knows its current position, the path with motion planning to the destination and its estimated arrival time. If the robot deviates from the move plan, the system will reschedule based on the current state. The experimental results show the great performance of our proposed method.
CO-EDITOR-IN-CHIEF:
ASSOCIATE EDITOR:
Nomenclature
A | = | Qn-1 in one step |
amax | = | maximum linear acceleration |
B | = | Qn in one step |
C | = | the intersection of the tangents of point A and B |
ct | = | the ratio that equals sdt/et |
cl | = | the lower threshold of ct |
Esti | = | the sensing error of the i-th particle at time t |
et | = | the minimum sensing error from all particles at time t |
l | = | half distance between two wheels |
O | = | the rotation center in one step |
P | = | the vertical projection of point B on the x-axis of point A |
Pi | = | the i-th node of the path |
Pia | = | split node of Pi |
Pib | = | meeting node of Pi |
P0 | = | the distance that the robot accelerates from 0 to vmax and then decelerate to 0 |
Q | = | the midpoint of two wheels |
Qn | = | the state of Q at specific discrete time n |
r | = | the turning radius of the inner wheel |
rk | = | maximum linear velocity ratio for turning |
rmax | = | maximum rotation radius |
sdt | = | the standard deviation of the particle array at time t |
sdt’ | = | the modified standard deviation from sdt |
ut1 and ut2 | = | control inputs |
vmax | = | maximum linear velocity |
Wt | = | the weight array one-to-one relative to the particle array at time t |
wr | = | radius of the wheel |
Xt | = | the particle array at time t |
Zt | = | distance data from the robot’s range sensors at time t |
αmax | = | maximum angular acceleration |
γ | = | gain to adjust the distribution range of particles |
θa | = | |
ωmax | = | maximum angular velocity |
ΔD | = | movement length of point P in one step |
ΔD’ | = | the Euclidean distance between points A and B |
ΔDl | = | movement length of the inner wheel in one step |
ΔDr | = | movement length of the outer wheel in one step |
Δx | = | the movement components on the X-axis in one step |
Δy | = | the movement components on the Y-axis in one step |
Δθ | = | the turning angle in one step |
Disclosure statement
No potential conflict of interest was reported by the authors.