Practical Navigation Experiment

During the robot’s operation, it may detect subjects that are unrecognized and therefore need to be authenticated to confirm whether they are permitted to be present in the robot’s environment or not. For this reason this experiment was conducted.

This experiment is important to test the potential of the system in approaching and/or chasing subjects that are not authorized within the robot’s environment.

4.5.1.1 Objective

The objective of this experiment is to test the operation and accuracy of the tracking and navigation system of the robot in different situations. This is important as it will highlight potential issues that may need attention, especially in dynamic and volatile environments.

4.5.1.2 Description

The experiment starts off by allowing a subject to appear within the robot’s environment. The robot is programmed to navigate to this person as would normally happen when an unrecognized and unverified subject is encountered by the robot.

First, the robot’s motion and face detection cameras would locate the subject’s initial location relative to the robot. The location information is passed onto the tracking camera which tracks the motion and change in subject location. Next the most recent location is used to determine the best approach path. Finally the robot starts to approach the subject based on the latest location information.

If an obstacle is encountered the robot will avoid it by adjusting its path. The tracking camera adjusts its direction to keep the subject within sight, once the obstacle is avoided; the robot adjusts its path to continue to approach the subject.

The subject is allowed to move about whether to approach the robot or to try to keep away from it, or even escape through the lab’s door. This is done to emulate real-life situations in which a subject may respond by cooperating or holding back from authenticating him/herself. Figure 4.19 shows the robot while navigating a narrow corridor to approach a subject during the experiment.

(a) (b)

(c) (d)

4.5.1.3 Setting

The experiment was conducted in the P08 L02 research lab of the Faculty of Electrical Engineering in UTM. The lab contains tables, chairs, as well as many other different objects. Figure 4.20 shows the experiment environment setting.

Figure 4.20: The lab environment in which the experiment was conducted

A wide angle camera of 120 degrees and a 640x480 resolution was used to locate and track the subject. The camera is positioned at a height of approximately 1.5 meters. The camera height can be adjusted to accommodate different requirements and/or work environments. Positioning the camera at this height allows the detection of short as well as tall humans. Figure 4.21 shows the robot used in the experiment.

The robot was equipped with two ultrasonic sensors tilted inwards at approximately 7 degrees to facilitate detection of different types of obstacles. Kindly refer to the obstacle avoidance section for details regarding the ultrasonic sensors used.

(a) The camera at the top (b) The equipment at the base

Figure 4.21: The robot used in the experiment

During the experiment, the subjects were allowed to move about (be dynamic) to keep the robot from reaching them. The subjects were also allowed to move the obstacles to test the robots maneuverability and change of path capability.

4.5.1.4 Results

The experiment was conducted in lab P08-02 in building P08 of the electrical engineering department of UTM. Several experiments were conducted to test the robot’s ability to detect subjects and approach them while avoiding obstacles. Table 4.11 shows the results that were gathered:

Camera

Ultrasonic Sensors

Table 4.11: The results from the navigation experiment

Action Completed

Successfully Notes

Navigating towards

a static subject Yes

The robot turns around to face the subject and then continues to move towards him/her.

Navigating towards

a dynamic subject Yes

The robot detects the subject initial position, and turns around towards him/her. The robot then continues to move towards the subject. When the subject changes his/her location during the robot’s approach, the robot changes its course accordingly.

Navigating towards a static subject with an obstacle on the way

Yes

The robot starts to move towards the subject, when the robot encounters the obstacle; it avoids it by changing its direction. When the robot has cleared from the obstacle, it turns around towards the subject to continue its approach. Navigating towards a dynamic subject with an obstacle on the way Yes

The robot starts by detecting the subject’s initial position and starts moving towards him/her. When the robot encounters the obstacle on the way, it avoids it by changing its movement direction. Since the robot is equipped with a wide angle camera, it can locate the position of the subject after clearing the obstacle and turns around towards the subject to continue its approach.

4.5.1.5 Discussion and Analysis

The results gathered from the experiment give several indications:

1. The robot is capable of detecting a subject from other object in the background.

3. The robot is able to navigate around obstacles while remaining on the same task of approaching the subject.

4. Dynamic subject relocation does not affect the robot’s capability of tracking and continuous approach of the subject.

During the experiment, the following issues were also observed:

 Vibrations resulting from the robot’s motion limits or prevents the robots from detect faces in the video frame. Operation effectiveness is inversely proportional to robot speed.

 The robot had to move somewhat slowly. This was due to increased instability as the speed increased. Robot stability is inversely proportional to robot speed.

 The faster the robot moves, the more complicated it becomes to coordinate the programs in the laptop and Arduino (microcontroller). Component operation coordination is inversely proportional to robot speed.

 The robot can steer smoothly towards the subject according to his/her position with reference to the robot’s pose.

Summary

Many experiments were carried out to evaluate the different aspects relating to the security robot. Each of the robot’s abilities were tested to find out the effects of different parameters on their operation as well as their effectiveness in different situations.

The camera’s field of view experiment helped to discover the most practical camera to be used in the project which is capable of covering all 360 degrees with the fewest number of cameras. The motion detection experiments helped in finding the most suitable threshold to use to achieve the best balance between effectiveness and stability in different lighting conditions.

The face size experiment helped in finding an equation to determine the approximate distance of the subject from his/her face size as well as determining the furthest distance for obtaining a practically usable face image using a Full-HD camera rather in comparison to a camera with lower resolution such as normal HD or VGA. The face detection experiment helped in finding the parameters that achieved maximum operation speed while allowing a high accuracy of correct face detection.

The navigation experiment helped in coordinating the robot’s direction of motion with the angle of the camera that is tracking the subject’s location. This was done by applying proportionate voltage levels to the dc motors driving the wheel relative to the location of the subject and the presence of obstacles.

The obstacle distance detection experiment helped in determining the effectiveness of the sensors as well the used tilt angle, while the navigation experiment helped shed light on the challenges associated with detecting a human face on a moving robot while avoiding obstacles and calculating the required motion to approach that person.

The face recognition experiment helped in finding the operational differences between the different face recognizers as well as determine the best face image size to be used to achieve best performance in terms of speed and accuracy.

The mentioned experiment were not the only conducted experiments. many other minor experiment were also conducted during the project to test the operation of the equipment, synchronizing them, coordinating their operation and determine the best setting.

CHAPTER 5