|
Learning Parameters from Manual Task Assignments for Mobile Robots |
|
Seil An, Dong Jun Kwak, H. Jin Kim(Seoul National University, Republic of Korea) |
| For case where exist some threats in field of task assignment, threat avoidance is needed to be considered. Then both fast arrival and threat avoidance are used to form the performance measure. The policies of task assignment decide which factor is mostly considered between arrival time and threat avoidance. The purpose of this research is estimating the hidden policy from user’s manual task assignment. Using Naive-Bayes classification, we achieved satisfactory policy estimation for task assignment from testing several users. |
|
|
Kinematic Control Algorithms and Robust Controller Design for Rescue Robot |
|
Seongil Hong, Won Suk Lee, Youn Shik Kang, Yong Woon Park(Agency for Defense Development, Republic of Korea) |
| This paper introduces the Korean rescue robot and presents the kinematic and dynamic control method. The
mission of the rescue robot is to move and lift patients or soldiers with impaired mobility for the rescue and assistance
in the battlefields, hospitals, hazardous and disastrous environments. |
|
|
Optimal Spline-based RRT Path Planning Using Probabilistic Map |
|
Kwangjin Yang(Korea Air Force Academy, Republic of Korea), Seng Keat Gan(The University of Sydney, Australia), Jinwook Huh, Sanghyun Joo(Agency for Defense Development, Republic of Korea) |
| This paper presents an optimal path planning algorithm that adopts the rapidly-exploring random tree(RRT) as a path planner. The RRT generates a valid path quickly, but it does not have the ability to control the quality of the path. In this paper, the nonholonomic path planning and the optimal path planning are tackled simultaneously within the RRT framework. The spline-based RRT algorithm generates a feasible path satisfying the differential constraints, and the RRT∗ strategy is incorporated to guarantee the optimal solution. |
|
|
Closed Loop-based extrinsic calibration of multi-modal sensors |
|
Sungdae Sim, Kiho Kwak, Jun Kim, Sang Hyun Joo(Agency for Defense Development, Republic of Korea) |
| We propose a new extrinsic calibration algorithm using closed-loop constraints for
multi-modal sensor configuration. The extrinsic calibration parameters are estimated by minimizing the distance between
corresponding features projected onto the image plane. We conduct several experiments to evaluate the performance of
our approach, such as comparison of the RMS distance of the ground truth and the projected points, and comparison
between the independent sensor pair and our approach. |
|
|
Probabilistic Traversability Map Building for Autonomous Navigation |
|
Juil Sock, Kiho Kwak, Jihong Min, Yong-Woon Park(Agency for Defence Development, Republic of Korea) |
| For the successful navigation of an unmanned vehicle, it is important to determine the traversable area reliably. In this paper, we address problem of building a traversability map for unmanned ground vehicle. Map building consists of two main parts; first part is the fusion of terrain recognition map and terrain model map. Second part is sequential update of the traversablity map. Maps are built from each sensors separately and fusion takes place to combine the maps. The resulting map is added to the existing map for update. The algorithm is implemented to an unmanned ground vehicle. |
|
|
Efficient 3D terrain mapping based on normal distribution transform grid |
|
Hyun Jun Na, Yungeun Choe, Myung Jin Chung(KAIST, Republic of Korea) |
| semantic map can be used for navigation, exploration and path planning. Among the various information ‘traversability’ is important before everything. Because only if traversability information is provided in advance, various high-level robot’s tasks are possible. The proposed efficient 3d terrain mapping method use NDT grid. And by using NDT grid’s covariance matrix, we define traversability using three different factors ‘slope’, ‘flatness’, ‘reliability’. Experimental results show by our proposed method traversability can be analyzed efficiently, and accurately. |
|