2015 Fiscal Year Research-status Report
動力学モデルに基づく球型転がりロボットの運動計画とその制御戦略
| Project/Area Number |
15K05900
|
|---|
| Research Institution | Kyushu University |
|---|
Principal Investigator |
M・M SVININ 九州大学, 工学(系)研究科(研究院), 教授 (90274125)
|
|---|
| Project Period (FY) |
2015-04-01 – 2018-03-31
|
| Keywords | ロボティクス / 非ホロノミックシステム / モーションプランニング / 最適制御 / 拘束運動 |
|---|
| Outline of Annual Research Achievements |
In the first year, we have developed the exact mathematical models of rolling robots with two different driving principles. The first principle is based the conservation of the angular momentum of the rolling robot and the second one is based the change of the center of mass of the robot. These driving principles can be realized by actuating a spherical shell with symmetrically placed orthogonal rotors (first principle), or with a pendulum mounted at the center of the spherical shell (second principle). The mathematical models developed for such robots take into account the dynamics of the driving mechanisms and nonholonomic constraints imposed on the motion of the rolling vehicle.
The controllability of the robots with minimal number of actuators has also been established. It has been shown that the decoupling of the motion planning to the kinematic and dynamic level for the rotors-actuated robot is impossible. The driftless control system describing motion of this robot is not differentially flat and not nilpotent. To steer the robot to a desired configuration, an iterative motion planning strategy, based on the nilpotent approximation, has been developed. The feasibility of this motion planning strategy has been verified under simulation. For the pendulum-actuated robot, the decoupling of motion planning is realizable, and the dynamic model can be reduced to two equation describing the system behavior in projection onto the vertical plane tangent to the contact path on the contact plane. This approach to motion planning has been successfully tested under simulation.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The results of our research have presented at several top-level international robotics conferences (IEEE ICRA and IROS conventions) and also at the important domestic conferences (annual RSJ and Robomec meetings). The key points and features of our research have been communicated in the robotics research community and a positive feedback and constructive suggestions have been received. Based on the opinions collected from top researchers, we can proceed with our research plan without considerable modifications and delays.
|
| Strategy for Future Research Activity |
The remaining part of our project combines theoretical and experimental investigations. In the theoretical part, we will use methods of geometric control theory in order to construct motion planning strategies and control algorithms for tracking desired trajectories. Here, we plan to formulate the optimal control and the tracking problems over the configuration manifold, explore the qualitative structure of its solutions, and come up with practical numerical algorithms and suggestions for online implementation of motion planning strategies. In the experimental part, we plan to design spherical rolling robots for the practical evaluation of motion control programs. Experimental work is very important part of our project since all the theoretical ideas about feasible control strategies should be verified under experiments. It is our intention to prototype the two rolling robots employing different driving principles. To improve the propulsion efficiency, a careful design of the transmission system between the tandem rotors with reconfigurable gear ratios will be undertaken. In the minimal configuration the robots will include a gyro sensor and an on-board computer system. The online implementation of the motion planning strategies and feedback control algorithms will be tested under experiments. It is very important at this stage to integrate different sensing schemes providing accurate measurements of the angular velocity of the rolling carrier and its position and orientation in the configuration space.
|
| Causes of Carryover |
This research combines both theoretical and experimental work. In the first year, we concentrated mainly on the theoretical developments of mathematical models, motion planning algorithms, and preliminary design considerations. Thus, the experimental part (the design of rolling robots) was in large shifted to the second year, and this was reflected in the actual spending. In the second year, we will continue the theoretical work on motion planning and tracking control and will come to actual prototyping of rolling robots with minimal number of actuators.
|
| Expenditure Plan for Carryover Budget |
In the experimental part, we plan to purchase basic components of the rolling robots which, in the minimal configuration, include four motors, two gyrosensors, and special spherical shells. Also, we plan to purchase a computer to control the robots, including the low level control software for programming the drivers. In the theoretical part of this research, we plan to purchase optimal control software tools to verify the correctness of theoretical constructions. In addition to the above mentioned, we also plan to discuss and disseminate our results at the top-level international and domestic robotics conferences.
|
