Room Searching Performance Evaluation for the JagaBotTM Indoor Surveillance Robot

This paper reports the evaluation of the performance of room searching capability for the JagaBotTM Indoor Surveillance Robot. The ultimate objective of the JagaBotTM is to be applied as an event actor, inspecting the secured environment automatically and feeding the dynamic view of the surrounding to remote user. It can also be used to communicate with persons inside the monitored area. A web service based instruction panel was used to command the JagaBotTM to designated rooms. The JagaBotTM then navigates itself to the room automatically, scanning for the QR code marker attached to room door, tracking a designated trail of lines through its QR code and line tracking camera. The result of this room’s searching procedure shows that the JagaBotTM achieved its objective of correct room finding in favorable time. A 100% correct search result was obtained with an average velocity of 0.1748 m/s under the current setting.


Introduction
Automated Ground Vehicle (AGV) type robot has been used in many applications including surveillance.The Smart Engineering System Research Group (SESRG) has successfully designed an AGV type robot, called the JagaBot  , for surveillance application.The robot is tasked with monitoring the environment dynamically, going from room to room and also to act as an actor, one which is commanded remotely to inspect a particular room upon detection of an anomalous event from that room (for example, from Closed Circuit Television (CCTV) observation manually or automatically).
There are several AGV robot developed for surveillance system, for example [1] presented a prototypical multi-robot surveillance system that is able to monitor an outdoor environment autonomously and visual surveillance system mounted on the mobile robot as developed by [2].A navigation architecture for autonomous mobile robot was developed by [3] to run in environment based on stereo vision camera utilizing the Binocular Stereo Vision Based Obstacle Avoidance technique.A robot for intruder detection and surveillance task was developed by [4] and [5] demonstrated a surveillance robot capable of capturing and transmitting video on rough terrains.[6] has incorporated unique feature of surveillance robot, which is travelling capability on both land and water.

JagaBo TM Infrastructure
The JagaBot  infrastructure is shown in Fig. 1 (b).The web service based command mechanism of JagaBot  is shown in Fig. 1  It uses an Intel NUC mini PC as the main controller, connected with a web cam for Quick Response (QR) code reading, a downward pointing line tracking camera, and two forward pointing wireless Internet Protocol (IP) camera for monitoring purposes.It also has two main screen, the larger one is used to display the Graphical User Interface (GUI) and the remote telecommunication module while the smaller one is used as a control panel.A Teensy microcontroller act as a field controller, connecting the NUC Mini PC to the sensor and actuator and all the instrumentation circuit below.The JagaBot  is move by two Direct Current (DC) motor and powered by 2 12V sealed lead acid battery.

Objective of Study
The objective of the study is to establish the performance of the JagaBot  in achieving targeted room for inspection and to derive a mathematical model depicting the target to arrival time with JagaBot  current location.
Figure 2 shows JagaBot  path in detecting the desired room.JagaBot  will go through the red line track for straight movement.When the green line marker is detected, JagaBot will make a turn based on the side detected and scan the QR code provided.If the QR code gives correct room reading, JagaBot  will enter the room.

Experiment Design
From the X  starting point, the robot were assigned to move to any of the rooms (room 1 (D1), 2 (D2), 3 (D5), 4 (D3), 5 (D4), 7 (D7)) using the web service based command panel.The experiment were repeated for several times.The floor plan of the experiment location is shown in Fig. 3 below.
Figure 4 shows the sequence of JagaBot  movement.It started from identifying the command for the room to be searched.Then, JagaBot  will move forward and find the room markers.Once JagaBot  detect the marker, it will turned to identify the room using QR code displayed in front of the door.If the QR code shows the wrong room, JagaBot  will turn back and continuously move forward to find the correct room.If the identified QR code shows the correct room, JagaBot  will enter the room for inspection.We have also develop a model to estimate the arrival time (T ) to destination.

Results
Table 1 shows the JagaBot  performance evaluation.The time taken for each target is measured using Eq. 1.
Where, T  : Estimated time for JagaBotT  to reach target room.
T  : Time spent when scanning correct QR code (Measured experimetally).
T  : Time spent when scanning incorrect QR code (Measured experimetally).
T ℎ : Time taken to move in a straight line from X  (shifted by T  ) to targeted door Marker (Measured experimentally).N : Number of turn.T  : Pick up time required for the JagaBot  motor to ramp up to nominal speed.

Accuracy
Both T  and T  depends on N and the distance from starting point.The T  is estimated using Eq. 1 and measured experimentally (T  ).The results of the experiment shows that all room were identified and reached succesfully.The average   percentage of error between T  and T  is 3.14% only, which indicates that the model is in good agreement with the real system behaviour.

Speed of Detection and Reach
The average measured value of V ℎ (V ℎ = Distance / T ℎ ) is 0.1748 m/s.However, T  and T  depends on the room distance and number of verification it has to do before reaching the correct target.Figure 5 shows speed and time performance of JagaBot

Conclusion
The JagaBot  was proven to be able to detect all room using QR code accurately.Good agreement was obtained between predicted arrival time to target room (T  ) and the measured arrival time(T  ), with an average percentage of error of 3.14%.Using this model, it would be possible to accurately estimate the arrival time based on JagaBot  current location.In future, the JagaBot  will be integrated into our (a).The instruction panel can be replaced by desktop instruction panel or android device based instruction panel.However, the web service act as the main communication hub for JagaBot  .Several versions of the JagaBot was developed.The version shown in Fig. 1 (b), which is the JagaBot-J03, is the latest version of the JagaBot  [7].

Figure 5 :
Figure 5: Time taken by JagaBot to reach target room.