Research and Development of Farm Mobile Robot for Sloping Terrain to Promote Mountainous District Agriculture
Grant-in-Aid for Scientific Research (C)
|Allocation Type||Single-year Grants |
Agricultural information engineering
|Research Institution||Iwate University |
TORISU Ryo Iwate University, Faculty of Agriculture, Professor, 農学部, 教授 (70038264)
TAKEDA Junichi Iwate Univ., Faculty of Agriculture, Associate Professor, 農学部, 助教授 (80133908)
IMAE Jyo Osaka Prefecture Univ., Faculty of Engineering, Professor, 工学部, 教授 (30184807)
|Project Period (FY)
2005 – 2006
Completed (Fiscal Year 2006)
|Budget Amount *help
¥3,700,000 (Direct Cost: ¥3,700,000)
Fiscal Year 2006: ¥1,200,000 (Direct Cost: ¥1,200,000)
Fiscal Year 2005: ¥2,500,000 (Direct Cost: ¥2,500,000)
|Keywords||farm mobile robot for sloping terrain / Neural Network vehicle model / on-tracking control system / generalization / tractor-implement system / broadcaster / platooning system / robustness / AI車両モデル / ファージ車両モデル / platooning|
1)Realization of Platooning system
A two farm tractor platooning system was developed. The farm tractor platooning system is defined the system that a preceding tractor which operated by a human-operator is pursued by a following mobile robot tractor on the farm. We designed and reconstructed the system and did the pre-test last year (2006).
(1)Tractor platooning system experiment on flat terrain
The exreriments were executed on a flat ground. Its average inclination is 3 degrees. 'II-actor velocities were set about 50(canisec). The distance between preceding and following vehicle was maintained 5(m). Especially the kinematic vehicle model was applied in this platooning system.
(2)Tractor platooning system experiment on sloping terrain
The exreriments were executed on a sloping terrain. Its average inclination is 10 degrees. 'II-actor velocities were set about 50(cm/sec). The distance between preceding and following vehicle was maintained 5(m). In this test the neural network(NN) vehicle mo
del was applied in the platooning system.
The results of this platooning system indicated that following mobile robot tractor followed up the leading one successfully with mean and RMS lateral deviations of 2.2 and 8.4 cm, respectively on sloping terrain.
2)Expansion of neural network vehicle model and its application for mobile robot on sloping terrain.
A NN vehicle model was formulated and reconstructed for estimating vehicle behaviors on sloping terrain. A new training method combined with genetic algorithm and back propagation algorithm was used to train the NN vehicle model.
3)Realization of Tractor-Implement System and its application for mobile robot on sloping terrain.
A broadcaster was mounted on the Mitubishi tractor that was adapted from the original to the mobile robot. The system weight was controlled on three levels as tractor itself(T), tractor plus implement and tractor plus implement with addition of fertilizer weight. The NN vehicle model developed for single tractor system were applied for two tractor+implement(T+I) and tractor + implement+fertilizer weight(T+I+200). T,T+I and T+I+200 system were traveled along the contour line on the sloping terrain. Field tests showed that for rectilinear paths(contour lines), the average tracking was -4cm, with standard deviation of 6cm, and for the curved paths, the maximum lateral deviation and maximum orientation error were less than 8cm and 11 degree respectively at the completion of these paths. Less
Report (3 results)
Research Products (5 results)