Page 243 - The-5th-MCAIT2021-eProceeding

3. Experimental Results

3.1 Simulation duct environment

The visualization of sensory data, robot position, and the duct environment using Rviz tools is showing in
Fig. 2. The robot's odometry is contained linear, and orientation is observed using a terminal by echoing
/odom topic. A total of 6 minutes and 16 seconds of simulation time is obtained to complete the whole
navigation to all 6 points. Our observation shows the robot required about 20 seconds to initialize its position
onto the map, although the predefined origin has been defined.

Fig. 2. Navigation to multi-goals using RRT
3.2 Path planning

Fig. 3 shows the robot navigational position from origin to all 6 references in sequence number. It is
observed that the trajectory movement of the robot is not too smooth using an existing algorithm. However,
all points are achieved with distance coverage of 51.6 meters and duration as presented in Table 1. It is
observed that simple path planning achieves every goal faster than RRT path planning. This is because RRT
path planning navigation uses a 2D Navigation arrow in SLAM for every goal, while simple path planning
uses autonomous navigation. Most existing algorithm are used using Turtlebot Indigo which is suitable only
in Ubuntu 14.04. Thus, we make some changes that suit Turtlebot3 in ROS Kinetic. The experiment was
conducted using Ubuntu 16.04 and Turtlebot3 in ROS Kinetic. Gazebo and Rviz are the platforms that are
used for simulation of the two different path planers to see their performance. To compare the performance of
the two path planning strategies, we measure the duration of every goal.

Goal 1 Goal 2

Goal 3 Goal 4

Origin
Goal 6
Goal 5

Fig. 3. Navigation of duct robotic based on six reference points.

E- Proceedings of The 5th International Multi-Conference on Artificial Intelligence Technology (MCAIT 2021) [228]
Artificial Intelligence in the 4th Industrial Revolution