behavior on uneven surfaces, and pruning.

Key words Pruning robot · Climbing robot

1 Introduction The timber industry in Japan has gone into decline because the price of timber is falling and forestry workers are aging rapidly. This has caused the dilapidation of forests, resulting in landslides following heavy rainfall and the dissolution of mountain village society. However, a pruned tree in a suitably trimmed state is worth money because its lumber has a beautiful surface with well-formed annual growth rings. The development of a pruning robot is important for the creation of sustainable forest management. The research and development of a pruning robot 1–5 has been rare, and Received and accepted: February 25, 2011

S. Ueki (*)

Department of Mechanical Engineering, Toyota National Colleges of

Technology, 2-1 Eiseicho, Toyota, Aichi 471-8525, Japan

e-mail: s_ueki@toyota-ct.ac.jp

H. Kawasaki · K. Koganemaru

Department of Human and Information Systems Engineering, Gifu

University, Gifu, Japan

Y. Ishigure

Marutomi Seikou Co. Ltd., Seki, Japan

Y. Mori

Hashima Karyuu Kougyou Ltd., Gifu, Japan

This work was presented in part at the 16th International Symposium

on Artifi cial Life and Robotics, Oita, Japan, January 27–29, 2011

87

the batteries. The center of mass was located with a margin of error, because the friction coeffi cient is unclear and the position of the center of mass may be moved by disturbance. For example, the robot will be tilted when it climbs up an uneven surface. In Fig. 2a, the center of mass was located with parameters H = 0.3 m and W = 0.22 m, where H is the distance between the upper side wheel and the lower side wheel, and W is the distance between the surface of the trunk and the center of mass, as shown in Fig. 3. The analysis shows that the robot is robust when D is 0.25 m, even if it is tilted about 0.1 rad. The controller is constructed using a CPU board which is equipped with a wireless LAN. The controller is able to communicate data/commands with a personal computer via the wireless LAN. Each wheel is controlled by a velocity PI control. A velocity feedback input through a high-pass filter is appended. By comparison with the 2nd prototype,8 the 3rd prototype is lightweight except for the controller and batteries. Also, the controller and the electrical source were located externally in the 2nd prototype. The 3rd prototype is also equipped with a wireless LAN and a chainsaw. Although details of the chainsaw are omitted here, an experiment was performed to show the cutting of a branch using the 3rd prototype.

3 Experiments

Three experiments were performed to evaluate the 3rd prototype. The 1st experiment was to evaluate its basic performance. The 2nd experiment was to evaluate its robustness on uneven surfaces. The 3rd experiment was to show whether the robot can prune a branch. All experiments were performed using a substitute tree indoors. The diameter of the substitute tree was 0.25 m. The frictional coeffi - cient of the substitute tree was about 0.4, which is less than that of a natural tree. To collect the experimental data, the motor current, the position of the robot, and the orientation of the robot were measured. The motor current was measured using shunt resistance. The position was measured by a 3-D position measurement device (OPTOTRAK, Northern Digital). The orientation was measured by a 3-D orientation sensor (InertiaCube2, InterSense).

Fig. 1. Tree climbing method using “BURINAWA”

Fig. 2. 3rd prototype of pruning robot. a Photo image. b CAD image reduction mechanism which has non-back-drivability. The steering angle of each wheel is also driven by the DC servomotor and the warm wheel reduction mechanism. Based on analysis,7–9 the center of mass was located outside the tree with the help of the weight of the controller and

Fig. 3. 3D fi gure of a pruning robot on a tree. a Side view. b Top view

88

3.1 Basic performance

A straight climbing experiment was performed to evaluate the robot’s basic performance. The desired speed of the four wheels was given by the trapezoidal profi le. The acceleration was 0.2 m/s2, and the speed was 0.2 m/s per 0.075 m of wheel radius. The experimental results are shown in Figs. 4, 5, and 6. Figure 4 shows the speed of the robot. The speed of each wheel was calculated from the values of the rotary encoder. The robot was able to climb at 0.2 m/s. Although there was a starting delay of about 0.5 s owing to the control law, this was not a problem. Figure 5 shows the distance moved. The “3D” value was measured by a 3D position measurement device, and the distance moved by each wheel was calculated from the value on the rotary encoder. In Fig. 5, we found three types of error: errors in the distance moved between each wheel and the 3D position measurement device (E1); error between wheel 1 (or 3) and wheel 2 (or 4) (E2); error between wheel 1 and wheel 3 (and error between wheel 2 and wheel 4) (E3). We considered two possible reasons for these errors. The fi rst was differences in the deformation of each wheel. The distance moved by each wheel was calculated as 0.075 m of the radius of the wheel. The wheel was composed of urethane and an inner tube which was deformed by the force acting on it. The deformation volume depended on the magnitude of the force. From a theoretical analysis,7–9 the magnitude of the force in the third prototype tended to be as follows. The normal force near the center of mass becomes larger than the force at the opposite side. Hence, Fn4 = Fn2 > Fn3 = Fn1 was considered, where Fni is magnitude of the normal force of wheel i. Both (E1) and (E2) can be explained in this way. We also considered that the reason for (E3) was slippage of the wheel on the trunk. Figure 6 shows the electric current in the wheel motors, which were measured by the shunt resistance. The theoretical analysis7–9 also showed that the tangential force on the lower side is larger than that on the upper side. Figure 6 tends toward the theoretical analysis.

3.2 Behavior on uneven surfaces

To use the robot safely, it must be robust on an uneven tree trunk. There will always be bumps caused by the growth of the remnants of a pruned branch. Therefore, a straight climbing experiment was performed to evaluate the robustness of the pruning robot for bumps on trunk. This experiment was performed on a substitute bump. The bump was made of ABS plastics, and was larger than a natural bump.The desired speed of the four wheels was given by a trapezoidal profile. The acceleration was 0.2 m/s2 and the speed was 0.2 m/s for every 0.075 m of the radius of the wheel. The experimental results are shown in Fig. 7, which shows the trajectories of angles 1 and 2 (see also Fig. 2b). Angle 2 rotated toward the plus direction in all cases, indicating that the control box was rising. This means that the center of mass moved toward the tree. The center of mass also moved toward the tree when angle 1 rotated toward the plus direction. This means that there is a decrease in the friction force keeping the robot on the tree. However, the electric currents in wheels 2 and 4 were larger than the continuous current in the experiment. Therefore, there was no danger of the robot falling down. Moreover, these angles returned to their former orientation, even though both angles 1 and 2 had changed when a wheel went over the bump. These results show the good robustness of the robot.

3.3 Pruning experiment An experiment was carried out to discover whether the 3rd prototype could prune a branch. An attached chainsaw was driven by a DC motor with a 24-V battery. The robot climbed the tree spirally at a speed of 0.03 m/s. The diameter of the target branch was 0.01 m.

Fig. 4. Climbing speed

Fig. 5. Climbing distance

Fig. 6. Electric current of each wheel

Fig. 7. Roll angle and pitch angle in each case. a Wheel 1 goes over the

bump, b Wheel 2 goes over the bump, c Wheel 3 goes over the bump,d Wheel 4 goes over the bump

Fig. 8. Pruning experiment with the pruning robot The experimental scene is shown in Fig. 8. In this experiment, the branch was cut off leaving only a short remnant which was less than 0.005 m, and the trunk was not injured.

4 Conclusion

The developmental progress of a timberjack-like pruning robot has been described, focusing on straight climbing, its behavior on an uneven surface, and pruning a branch. The straight climbing experiment showed that the 3rd prototype gave a good basic performance. The result of the climbing experiment on an uneven surface showed good robustness for bumps, because most bumps on real trees are smaller than the experimental bump. Moreover, the pruning experiment

also showed that the 3rd prototype can prune a branch from a tree.In future work, we hope to test the robot in a real environment,

and try to make some further improvements.

References

1. Takeuchi M, et al (2009) Development of street tree climbing robot

WOODY-2 (in Japanese). Proceedings of Robomec 2009, 1A2–D07

2. Kushihashi Y, et al (2006) Development of structure of measuring

grasping power to control simplifi cation of tree, climbing and

pruning robot Woody-1 (in Japanese). Proceedings of the 2006

JSME Conference on Robotics and Mechatronics

3. Suga Y, et al (2006) Development of tree-climbing and pruning

robot WOODY. Actuator arrangement on the end of arms for

revolving motion (in Japanese). Proceedings of SI2006, pp

1267–1268

4. Yokoyama T, Kumagai K, Arai Y, et al (2006) Performance evaluation

of branches map building system for pruning robot (in Japanese).

Proceedings of the 2006 JSME Conference on Robotics and

Mechatronics

5. Yamada T, Maeda K, Sakaida Y, et al (2005) Study on a pruning

system using robots: development of prototype units for robots (in

Japanese). Proceedings of the 2005 JSME Conference on Robotics

and Mechatronics

6. Seirei Industry. http://www.seirei.com/products/fore/ab232r/ab232r.

html. Accessed May 2011

7. Kawasaki H, Murakami S, Kachi H, et al (2008) Analysis and experiment

of novel climbing method. Proceedings of the SICE Annual

Conference 2008, pp 160–163

8. Kawasaki H, Murakami S, Koganemaru K, et al (2010) Development

of a pruning robot with the use of its own weight. Proceedings of

Clawar 2010, pp 455–463

9. Kato T, Koganemaru K, Tanaka A, et al (2010) Development of a

pruning robot with the use of its own weight (in Japanese). Proceedings

of RSJ2010, Nagoya