This paper presents a method to precisely control an Autonomous Mobile Robot(AMR) vehicle driven by two in-wheel BLDC motors. The proposed method is based on the X-Y coordinate axis and estimates the driving angle.
In general, the controller in Automated Guided Vehicle(AGV) or Autonomous Mobile Robot(AMR) is divided into two categories : higher and lower level. The higher-level controller manages the entire control process based on the condition of the surrounding environment and the later operates the motor to move the vehicle. In order to control the position of an AGV or AMR, it is important to know the location and position of the vehicle. In most conventional systems, the higher-level controller is used to control the position and the lower-level one receives the resulting speed reference command and runs the motor.
The higher-level controller in current AMR vehicles uses sensors or devices to identify the locations such as GPS, line tracer, RFID, camera, Lidar, etc. However, these added devices can create various limitations and problems. For example, GPS-based systems are difficult to use indoors due to poor reception of satellite signals and line tracer systems can only move on a pre-determined route. The installation distance for each RFID has to be close to increase accuracy and the systems based on machine vision require a separate device for image processing. Therefore, it can be concluded that the precision of the higher-level controller is greatly affected by the performance of the position-detecting device. Additionally, two or more sensors can be used together to increase precision, but this complicates the systems and increases the cost. To solve this problem, a control method based on the lower-lever controller is proposed in this dissertation by estimating the vehicle driving angle using an IMU (Inertial Measurement Unit). Two kinds of movement paths consisting of straight and curved routes are considered and various problems which occur during driving are observed.
The control of vehicle basically includes acceleration and deceleration modes. However, when conventional method is applied, there are errors due to the overshoot of speed and inertia caused by the low resolution of Hall sensors in BLDC motors in the low-speed range. In the straight-path case, these errors can cause the vehicle to deviate from the set path and consequently fail to reach the target point, even though the total distance travelled is the same as the given command. On the other hand, in the curved path case, the error becomes larger since the commanded driving angle and moving distance are varied over time for the two motors.
In this dissertation, a method to set the acceleration and deceleration patters for both straight and curved paths based on the X-Y coordinate axis is proposed. In addition, the compensation method for angle, position, and slip errors during driving by using the driving angle estimation data from the IMU is also presented. The performance of the proposed method was verified through various tests with a vehicle-type AMR with two in-wheel BLDC motors. It can be confirmed that the precision of position control was improved using the proposed method.