Indoor Navigation System for Mobile Robot using Wireless Sensor Network Wooyong Lee, Minkyu Kim, Wonhee Yee, and Dooseop Eom Department of Electronics and Computer Engineering, Korea University #1-5 ga, Anam-dong, Sungbuk-gu, Seoul 136-713, Korea Tel: +82-2-3290-3802, Fax: +82-2-3290-3895 E-mail: {waryong, mensama, wangpepe}@final.korea.ac.kr, [email protected]

Abstract: We describe an algorithm for indoor mobile robot navigation using wireless sensor network. In our navigation system, a robot can navigate autonomously without the need for a map, a compass, or GPS. We use only several sensor nodes with ultrasonic sensors. Sensor nodes are deployed in an indoor environment and each sensor node has routing paths to all available destination-nodes through flooding. Measuring the distance from one sensor node after another along the routing path, the mobile robot moves toward the direction that shorten the distance and finally comes to reach the destination. The validity of the proposed algorithm has been proved by many experiments.

1. Introduction Navigation is a significant problem in mobile robotics, ubiquitous computing, and URC 1 . So many solutions about mobile robot navigation have been proposed. But most rely on navigating using a pre-specified map in [1], [2] then an applicable environment is limited. Some algorithms combining mobile robots with static sensor networks are proposed and maps are not needed. But in [3], error range is relatively large because distance measurement depends on RF RSS (Received Signal Strength) holding poor distance resolution. And in [4], a cost of the system is high because additional devices such as GPS module and digital compass are used. In this paper, we propose an algorithm for indoor robot navigation using wireless sensor network with ultrasonic sensors. Without the need for a map, a compass or GPS, automatically a robot can navigate merely measuring the distance from a sensor node. Our approach without localization technique is so simple that this algorithm is easy to be embedded in hardware devices. In our navigation system, for measuring the distance between the robot and each sensor node more accurately, ultrasonic sensors are used. As shown in Fig. 1, the robot is composed of two stepmotors, batteries, operating circuits, and one sensor node2. And other sensor nodes are attached to the ceiling (Fig. 2) and form an ad-hoc sensor network. Sensor nodes which we have used were specified in Table 1.

Fig. 1. Mobile robot with sensor node and ultrasound receiver. 1 2

URC(Ubiquitous Robotic Companion) Hereafter, we regard this sensor node as the robot.

Fig. 2. Sensor node attached to the ceiling with ultrasound transmitter. Table 1. Sensor node specification.

2. Proposed Algorithm Navigation Environment: In our navigation system for mobile robot, sensor nodes act as signposts and according to their routing paths they inform the robot of the next node to pass through. And the robot follows each sensor node along the routing path and interchange distance information with each sensor node. For measuring distance more accurately, ultrasonic sensors are used and sensor nodes are attached to the ceiling. Fig. 3(a) indicates the valid coverage of the ultrasonic sensor. If the height of each sensor node is 2.7m, the radius of the region an ultrasound reaches is about 1.5m and distances between sensor node and one hop neighbors should be less than 1.5m. Navigation Algorithm: The main procedure of this navigation system is as follows. First, those sensor nodes attached to the ceiling maintain routing tables through flooding [5]. The routing table of each sensor node just lists all available destinations and the next hop to each, not hop count. Second, the robot finds the closest node by measuring distances from the sensor nodes. Each node on the ceiling transmits RF signal and ultrasonic pulse simultaneously, and if the robot receives these signals then it estimates distance using the time difference in RF and ultrasonic signal propagation. Third, the robot moves from the closest node to the next node (hop) and finally to the goal passing

.Fig. 3(a) Valid coverage of the ultrasonic sensor (b) Virtual distance x d 2  h2 (c) The movement of the robot within sensor network through the point just under each node along the routing path. Fig. 3 shows these phases more specifically. In Fig. 3(b), if the robot measures the distance d, the virtual distance x is calculated because the height h is constant. In Fig. 3(c), initially the robot is located at A and should move to C. But the robot can’t know the direction to C, so it moves forward to initially set direction and arrives at B. At this point, with virtual distance-values a and b it can compute the angle T as shown in (1), (2).

cos(S  T )

T

b2  s 2  a 2 2bs § a2  b2  s 2 · ¸¸ arccos¨¨ 2bs © ¹  cosT

Acknowledgements This research was supported by University IT Research Center Project. Table 2. Experimental results. (Distance Error : distance from the robot to the point just under the goal node)

(1) (2)

Because the robot runs by two step-motors, it knows the shifting distance s and can change its direction precisely. Periodically estimating the virtual distance, it runs forward until the virtual distance is close to zero. If arriving at the vicinity of C, it interacts with the next node and repeats former procedures and finally comes to reach the destination.

3. Experiments We implemented the algorithms described above in the environment that two sensor nodes were attached to the ceiling with 1.5m apart from each other and the robot would move from any region of one node to the other node. And we measured a distance error after the robot changed its direction one time. The experimental results are shown in Table 2. We have not yet tested more complete cases that many sensor nodes were deployed, but this system is less dependent on scalability because an error could not be accumulated.

4. Conclusion In our navigation system, the robot can navigate autonomously with aid of a sensor network and additional entity such as middleware-server is not needed. So this system can be constructed at low cost and if temperature sensors are mounted to nodes it is also possible for the robot to avoid danger (hot) regions by changing the routing path.

References [1] D. Kortenkamp and T. Weymouth, “Topological mapping for mobile robots using a combination of sonar and vision sensing,” in Proceedings of the AAAI, 1994. [2] R. Simmons and S. Koenig, “Probabilistic robot navigation in partially observable environments,” in Proceedings of the Int. Joint Conf. on Artificial Intelligence, 1995. [3] Maxim A. Batalin, Gaurav S. Sukhatme and Myron Hattig, “Mobile Robot Navigation using a Sensor Network,” in IEEE Int. Conf. on Robotics and Automation, 2003. [4] Tae-Kyung Moon and Tae-Yong Kuc, “An Integrated Intelligent Control Architecture for Mobile Robot Navigation within Sensor Network Environment,” in Proceedings IEEE/RSJ Int. Conf. on Intelligent Robots and Systems, 2004. [5] C. E. Perkins, Ad Hoc Networking, Addison Wesley, pp. 173219, 2001.