• Author / Uploaded
• Waasi Khan

Citation preview

Proceedings of the 2011 IEEE International Conference on Mechatronics and Automation August 7 - 10, Beijing, China

A Novel Shrimp Rover-based Mobile Robot for Monitoring Tunnel Power Cables Songyi Dian, Tao Liu, Yan Liang and Mengyu Liang

Wei Zhen

School of Electrical Engineering and Information Technology Sichuan University Chengdu, Sichuan Province, 610065, China [email protected]

Sichuan Electric Power Test & Research Institute Sichuan Electric Power Corporation Chengdu, Sichuan Province, 610072, China [email protected] distributed sensor network, this method is much cheaper and lower risk on execution. Third, the robot can gather information about aging status of the cables more precisely while the cables charged. Prior work to make the robot inspection method widely used in practice is to design a robot which can move along the tunnel cables stably. One of the challenging issues in the monitoring robots design is their ability to handle unstructured working environment. There are two main kinds of obstacles during the robot travelling along cables. One is the fastening clip. The other is the cable joint which connect two dividual cables being of limited length to yield more long distance. Some slopes will not be avoided due to complex distribution and field installation of cables. Therefore, a flexible robotics system that can sustain difficult conditions with a dependable control system is essential for the tunnel power monitoring. One robotic platform for monitoring underground power cable systems has been developed by University of Washington [4, 5]. It is a unique segmented configuration which contains power drive module, sensor array and additional segment. But this kind of robot has not enough ability to climb those accessories on cables. It is hard to use in actual environment. In this paper, a novel wheeled mobile robot based on shrimp rover is presented. The Shrimp rover robot has somewhat similarities to the sea creature “shrimp” in motion principle. Using rhombus configuration, there are two bogies assembled on right and left sides of the robot’s body, respectively. Each bogie has two wheels. In the front and the rear of the rover robot, there are two steering wheels named the front wheel and the rear wheel, respectively.The front wheel has a spring suspension to guarantee optimal ground contact of all wheels at any time [6]. However the traditional shrimp design could not be competent for the task of tunnel cable monitoring, some improvement have made on it so that it can be used for this task. The first advantage of this structure is the all terrain locomotion which allows travelling along cables and climbing obstacles. The second advantage is that, just the control methods for wheel-driven motors and balance slides are necessary for robot locomotion.

Abstract – Accurate, real-time information about the aging status of tunnel power cables can reduce the risk of urban powergrid breakdown effectively and avoid huge economic losses due to kinds of failures of power lines. Because of tunnel power cables being in quite rigorous environmental conditions, it is hard to get the aging status information by people themselves or normal robots. To cope with this challenge, we propose a novel wheeled mobile robot for unstructured environment based on existing shrimp rovers. It can climb over the obstacles on power cables thanks to its special mechanical structure and the balance slide which makes the robot keep balance constantly while tracking along tunnel power cables. Index Terms – Power cables monitoring; Mobile robot; Locomotion; Balance Control;

I. INTRODUCTION From the 1980s, the tunnel power cables have become the main way to transmit electric power rather than overhead power transition lines in cities. Economically effective maintenance and monitoring of power systems to ensure high quality and reliability of electric power supplied to customers is becoming one of the most significant tasks of today’s power industry. Nowadays the preventive maintenance is widely used in status monitoring of tunnel power cables. This method is to inspect the tunnel power cables by the workers while cutting the power or replace the cables according to scheduled and corrective maintenance procedures. However it will increase the difficulty of working arrangement. Moreover considering periodical replacement of cables, it is possible to replace prematurely a cable being in usable condition. According to previous studies related with anticipated cable replacement in scheduled maintenance, only 1/3 of these cables need to be substituted [1, 2]. On the other hand, with previous diagnosis, it would be possible to reduce the cost with cables and services. Some methods for monitoring and preventive diagnostic are based on distributed sensor network or very specialized technical inspection. According to [3], the main drawbacks of such methods are high cost, low accuracy and risk on execution. With the continuing technology improvement of autonomous robots, it is possible to develop a robot which can inspect underground tunnel power distribution cables. First, compared with the preventive maintenance, using the robot can reduce the cost of power cutting, and ensure the security of monitoring workers. Second, compared with the

978-1-4244-8115-6/11/$26.00 ©2011 IEEE

II. DESIGN OF ROBOT LOCOMOTION This novel shrimp rover illustrated in Fig. 1 is a robot with six wheels correspondingly driven by six motors. It has one front four-bar to climb over a certain height of obstacles

887

without any stability problem. The middle four wheels have parallelogram bogies which balance the wheel reaction forces during climbing. The single rear wheel is connected directly to the main body also driven by a motor to increase the climbing capacity. Balance pole which improves the robot’s stability on the cable is on the top of the robot’s body. As shown in Fig. 2, the parallel architecture of the bogies and the spring suspended fork provide a non-hypostatic configuration for the six motorized wheels while maintaining a high ground clearance. This insures maximum stability and adaptability as well as excellent climbing abilities. The robot is designed to keep all its six motorized wheels in contact with the ground on a convex ground.

Force Force

Fig.4 The front fork

z

A. The Bogies The bogies are the most important component of the rover. They provide the lateral stability during the travelling even on the rough terrain. To insure good adaptability of the bogie, it is necessary to set the pivot as low as possible and in the same time to keep a maximum ground clearance. As it is showed on Fig. 3 that compared with classical bogies, parallel bogies have a larger ground clearance which makes robot can climb higher obstacle [7].

z

B. The Front Fork The front fork designed as Fig. 4 has three functions: z It is possible that the front wheel can touch the cable all the time because of the spring.

When the robot encounters an obstacle, the horizontal force acting on the front wheel creates a torque around the instantaneous rotating center of front wheel. The four-bar mechanism design in the front wheel shows that the instant center is set under the horizontal line, and therefore causes the wheel to move up accordingly. When the front wheel is going up, spring is compressed and energy will be stored in the front wheel. Although other wheels are not in a good condition during climbing and they don’t touch the ground completely, this stored energy helps them move up easier.

C. The Six Wheels The friction between wheels and the cable provides the main propulsion to make robots travelling long cables and climbing over obstacles. General wheels have less friction due to their small contact area with cables. However, hourglassshaped wheels can improve the situation. They have been implemented in both front and rear wheel. And the halfhourglass-shaped multi-level wheels are used for four bogie wheels (Fig. 5). Using this kind of design brings three advantages: z Hourglass or half hourglass-shaped wheel has a larger contact area with cables which ensures enough friction for the robot to move over obstacles and function longer. z This design can enhance the robot’s stability on the cable. z The multi-level structure is to cope with obstacles on the cable especially for cable joints. The first level is designed following the cable’s diameter so that the wheel can fit better the cable. The second one has the same function with the first one, except that it is designed particularly for cable joints. The third level of the wheel improves the robot’s climbing ability. Thanks to this novel design of the driving wheels, all the wheels fit the cables very well. So the whole robot can travel around small curves which are unavoidable in monitoring task because of the rigorous environmental conditions and complex distribution of the cables if each wheel can steer a certain angle. Therefore, as illustrated in Fig. 6, the movable joints which affect robot’s steering can enhance the robot’s turning ability.

Fig. 1 Model of modified shrimp robot

Fig. 2 The robot on convex or concave line

Fig. 3 The bogie architecture

888

Fig.5 (a) The common wheel, (b) The hourglass-shaped wheel, (c) The halfhourglass-shaped multi-level wheel

Fig. 7 Motion principle of the balance slide

Two designs are shown in Fig. 9. The picture (a) presents an improper model. Because the front wheel moves forward while climbing as shown the path of EE’, which makes the whole robot back up, so the robot will get stuck. The picture (b) is a proper one which satisfies all the designing standards. 2) Containment interference conditions Containment interference conditions mean that the size of the obstacle should satisfy the requirement that no other components of the robot except the wheels should be contacted with the obstacle during striking an obstacle [9]. It indicates that if the obstacles were out of this limitation, the obstacle-crossing would fail. When the robot crosses long slope obstacles, it will conquer them successfully due to its special mechanical structure. Only rectangular obstacles may not meet the inclusion conditions. It can be found that the robot’s body is relatively lower than other parts as shown in Fig. 10. So the obstacles should not satisfy following conditions at same time:

Fig. 6 (a) The front steering joint, (b) The bogie steering joint, (c) The rear steering joint

D.

The Balance Slide The balance status while the robot tracking along tunnel power cables cannot be constantly maintained if only adjusted by the friction between the wheels and the cable. Hence, a slide is equipped to ensure the robot’s balance ability while walking along cables. As shownd in Fig. 7, the centre of gravity can be adjusted anytime by moving the slide on the balancing pole. The slide can be driven by linear motor to move right or left quickly and exactly. III. MECHANICAL ANALYSIS A. Kinematics Analysis 1) Designing Standards of the Front Fork The robot’s climbing behavior much relies on the front fork. A non-proper elbow size will bring failures during the obstacle climbing. Some designing standards are listed as below: z Front wheel proper going up while striking an obstacle z Front wheel proper range for rising and descending z Non-existence of death point that front wheel should not move forward while striking an obstacle in the mechanism. In the process of lifting the front wheel (Fig.8), CE rotates counter-clockwise relative to the AB with the center as F; and E upward velocity direction is perpendicular to the EF. In order to decrease xE gradually during this process, it is necessary to guarantee yEyF in coordinate XOY. However A and B are fixed, so yF is decreasing as the increasing of , meanwhile yE is larger and larger. So if yE=yF can be satisfied in the initial state, than yE will be always larger than yF [8].

Fig. 8 Analysis of the front fork design Cÿ C ÿ C

D ÿ n

C

D ÿ

B D

n

D A

B

Eÿ

A

Eÿ

E

E

a

b

Fig. 9 Basic condition of the robot for overcoming obstacles

889

L1

a r

b

e

H L

c

d

Fig. 10 Containment interference condition

L d L1  2 r H ter

(1) (2)

e

f

Fig. 11 Sequences showing the rover climbing a step

Height (mm)

3) Path of the Robot while Striking an Obstacle One of the most important requirements of this robot is the capacity of overcoming a stepped obstacle with a height of 65 mm which would hinder the robot when it is monitoring the cable. The Fig. 11 shows the main sequences of the robot climbing an obstacle. At beginning, the front wheel gets on the step with the suspension spring compressing. This relieves the load on the bogie wheels, and helps the bogie wheels up to the step. When the second bogie wheels climb up the step, the center of gravity reaches almost its final height. Finally, it is easy for rear wheel to get on the step as it is dragged by the other five wheels. As the two bogies are independent from each other, it is even possible to climb step if the robot is not approaching perpendicularly or only one bogie encounters an obstacle. Fig. 12 shows the path of the robot’s body during climbing over the step. The trajectory clearly demonstrates that the movement of the center of gravity of this mechanical structure is smooth. This key idea makes the system much better than other concepts.

Time (s) Fig. 12 Path of the robot’s body

IV. ANALYSIS AND SIMULATION OF BALANCE CONTROL The unique design of wheels has made the wheels fit the cables very well. But it is still possible to fall from the cable after long distance moving especially encountering some obstacles. In order to complete the inspection task smoothly, the balance of robot on the cable should be considered.

Dynamics Analysis 1) Necessary Torque of the Front Wheel The necessary torque of the front wheel is determined by many elements such as spring stiffness, front mechanism size and pre-compression of spring. Fig.13 shows the necessary torque of the front wheel, and the maximum torque which is less than the motor’s maximum torque occurs at 2.7 s when the first two bogie wheels are climbing up the obstacle. 2) Energy Based Analysis The springs keep the front wheel contacting with cables all the time. It can also store energy when the robot’s front wheel moves upward. And this part of energy will reduce the needed torque of first two bogie wheels when they are climbing over obstacles (Fig. 14). At the beginning of climbing behavior, the bogie wheels should overcome the change of spring, so the needed torque is a bit larger than the one without spring. When the first two bogie wheels are going up, the needed torque reduces significantly thanks to the released energy of the spring.

Torque (N/mm)

B

Time (s) Fig. 13 Necessary torque of the front wheel

890

Angle offset (º)

Torque (N/mm)

Time (s) Fig. 16 Result of the simulation

Time (s) Fig. 14 Needed torque of the bogie wheel in two case: with and without of the spring

Input of the control (N)

Because of the balance slide, the robot’s centroid can be adjusted by moving the slide on the balance pole when the robot’s position is not horizontal. For example, when the robot displaces toward left, the slide is moved to right of the balance pole by controller. This program uses the dynamic analysis software ADAMS and control simulation software MATLAB/SIMULINK together to build a balance control system. The description of the robot’s system can be exported by ADAMS; meanwhile MATLAB builds a control strategy based on the description (Fig. 15). The output of the robot system is the angle of vertical offset, and the input is the force acting on the slide along the balance pole. The definitions of output/input were done in ADMAS, and they will be imported to MATLAB with the description of the system together. Simulation condition is that robot is travelling on a cable with diameter of 94 mm, and the obstacle on the cable is a cylinder with diameter of 160 mm, high of 80 mm. The result of simulation is shown in Fig. 16 and 17. From the 2 s to the 7.5 s demonstrates the process of climbing the obstacle, and the largest angle offset occurs at the 6s when the rear wheel is climbing up the cylinder. Balance control is at work by moving the slide to the right and left throughout the whole process. The robot returns the state of equilibrium at the 12 s.

Time (s) Fig. 17 Input of the control

Some important points of simulation are listed as below: z The definition of system input and output in ADAMS must be correct. z The slide should have enough mass to adjust the robot’s centroid effectively. z It is necessary to consider the algebraic loops problems [10] during the simulation. V. FUTURE WORK AND DISCUSSION This robot’s mechanical design is completed, and the simulation on software is done as well. The results of simulation verify that the design of robot’s structure and control algorithms are reasonable. Because of its special design of locomotion, this rover has better adaptability than the robotic platform mentioned in [4, 5] which is much restricted by surrounded environmental conditions, especially for its stabilizers. Furthermore, the grade ability, typical obstacle climbing ability and attitude control make this modified shrimp robot be more eligible for monitoring tasks of tunnel power cables than other vehicles. The next step is to choose proper motions, materials, controller, and to develop a prototype. Then have it tested in laboratory and actual environment. A good distribution of wheel speeds and torques based on the information provided by the model and the sensors allows optimizing the robot’s motion. Hence, a better controller of the robot’s motion will reduce the wheel slippage, decrease the overall energy consumption and improve the robot’s climbing performance. Besides, in order

Fig. 15 Block diagram of the control system

891

[8] Cao Chongzhen, Zhang Jiliang , Wwang Fengqin, Zhao Chunyu, Zhuang Jialan, “Synthesis of the Head Mechanism for a New Shrimp-shaped Sixwheel Mobile Robot,” Journal of Shandong University of Science and Technology Natural Science, vol. 29, no. 1, pp. 58-61, Feb 2010. [9] Zhang Jiliang, “Study of the shrimp-shape six-wheel mobile robot and optimization to the compatibility of overcoming the obstacle for robot,” Shandong: Shandong University of Science and Technology, 2009. [10]Geng Hua, Yang Geng, “Algebraic loop problems in simulation of control systems and the methods to avoid it,” Electric Machines and Control, vol.10, no. 6, pp.632-635, 2006.

to have better disturbance restraint performance for the balance control, an optimal control strategy will be further considered. Sensors for power cable monitoring are also extremely important as they directly affect the gained information about cable’s status of insulation. It is necessary to ensure the robot’s moving ability while also leaving enough room for the installation of sensors. Then the multi-sensors signals fusion will be another challenge for this project. Monitoring tunnel power cable is a task which takes long time. So the energy consumption is another issue for this rover. However this aspect must be taken into account in every part of the system. Not only the power module must be powerful enough, but also the optimal trajectory algorithm, high-strength lightweight materials, and high coefficient of friction wheels must be considered. VI. CONCLUSION In this paper a novel Shrimp-wheeled robot for monitoring tunnel power cables is proposed. The results of simulation have shown that this rover has successfully coped with two challenges: z Based on a parallel architecture, which allows high ground clearance, this rover is able to overcome stepped obstacles passively and to move in rough terrain. z Thanks to the balance slide, it guarantees the balance ability of the robot while travelling along cables and climbing over obstacles. The excellent climbing capability enables this modified shrimp robot to overcome surmountable obstacle rather than to aviod them. It facilitates the robot to travel along lines successfully while performing status monitoring in mobile manner for tunnel power cables. Therefore, the robot discussed in this paper is a prefect candidate for tunnel power cable monitoring. REFERENCES [1] W. Reder and D. Flaten, “Reliability centered maintenance for distribution underground systems,” IEEE Power Engineering Society Summer Meeting, vol. 1, pp. 551 – 556, 2000. [2] P. Birkner, “Field experience with a conditional-based maintenance program ok 20-kv xlpe distribution system using irc-analisys,” IEEE Transactions on Power Delivery, vol. 19, no. 1, 2004., pp. 3 – 8, 2004. [3] R. Lyle, “Effect of testing parameters on the outcome of the accelerated cable life test,” IEEE Transactions on Power Delivery, vol. 3, no. 2, pp. 434 – 439, 1998 [4] B. Jiang, P. Stuart, M. Raymond, “Robotic platform for monitoring underground cable systems,” IEEE/PES Transmission and Distribution Conference and Exhibition, vol. 2, pp. 1105-1109, Oct 2002. [5] Bing Jiang, Alanson P. Sample, Ryan M. Wistort, and Alexander V. Mamishev, “Autonomous Robotic Monitoring of Underground Cable Systems”, Proceedings of the 12th International Conference on Advanced RoboticsΔJuly 2005 : 673 – 679 [6] Frat Barlas, Design of a Mars Rover Suspension Mechanism, Izmir: Izmir Institute of Technology, 2004. [7] Lamon, Thomas Estier, Michel Lauria, and Ralph Piguet, “Innovative design for wheeled locomotion in rough terrain,” Robotics and Autonomous Systems, vol. 40, no. 2, pp. 151-162, August 2002.

892