| PARENTE BURDELIS MAURICIO ALEXANDRE | D | 池田和司 小笠原司 柴田智広 |
|
発表題目：Modeling and estimating passive dynamics distributions in linearly solvable Markov decision processes
発表概要：Todorov has recently introduced a class of linearly-solvable Markov decision processes (LSMDPs) which greatly simplifies reinforcement learning. By attending specific conditions, the problem of choosing optimal actions (sequential decision making) can become linear, and the computational complexity involved in solving the Bellman equation is drastically reduced. One important feature of this framework is the introduction of the “passive dynamics” probability distribution, which corresponds to the behavior of the system in the absence of control actions. The purpose of the present work is to identify convenient models of the passive dynamics distribution in order to enable the application of the LSMDPs framework to reinforcement learning problems. We present results of experiments in a grid-world environment, which aim to recover the passive dynamics distribution from existing LSMDP models both by observing the behavior of the agent as well as observing incurred costs. We also discuss future steps to improve the estimation process and make it applicable to more realistic environments. | ||
| 的場俊亮 | D | 杉本謙二 小笠原司 野田賢 |
|
発表題目：局所同次システムを利用したロバスト有限時間整定制御則設計
発表概要：離散時間制御システムに対する有限時間整定制御がよく知られているが， 近年これとは異なる原理による非線形有限時間整定制御が注目を集めている． 非線形有限時間整定制御は，非線形時不変状態フィードバック制御則によって連続時間制御システムに対する有限時間整定性を実現するものであり， 実システムに対する制御性能の向上，とりわけ高速・高精度化が期待されている． 有限時間整定制御手法のひとつとして，中村らは制御Lyapunov関数(CLF)を用いた，パラメータ変化に対してロバスト性を保証する局所同次逆最適状態フィードバック制御則を提案している． この中村らの制御則は，有限時間整定制御以外の収束速度をもつ制御則設計も可能である上，制御則設計および実装の両面において計算量が少ないことが示されている． しかしながら，これまで実機実験による制御則の有効性は確認されていない． 本論文では，ロボットマニピュレータの位置決め制御問題を考える． 工作精度の向上のために産業用ロボットの制御性能の向上は強く求められている． ロボットマニピュレータの制御性能の向上，特に高精度化においては不連続に変化する摩擦力が大きな問題であるが． これは同時に中村らの制御則をロボット制御に適用する上での大きな障害にもなっている． そこで本論文では，まず不連続摩擦を考慮したロボットマニピュレータの位置決め制御問題に対し，中村らの有限時間整定制御則が有効であることを理論的に示し，中村らの有限時間整定制御則がその他の収束速度をもつ制御則と比較して高速かつ高精度な制御性能を持つことを実機実験により明らかにする．さらに，これまでに提案されている非線形有限時間整定制御則と比較して，中村らの制御則が高精度な位置決め制御を実現することを示す． 最後に本研究の今後の展望について述べる． | ||
| Mejia Ramon Francisco Pacquiao | D2 | 関浩之 湊小太郎 楫勇一 |
|
Title: Low-Density Parity Check Codes for High-Density 2D Barcodes
Abstract: Increasing the data density of two-dimensional barcode symbology is an important area of research in Automatic Identification and Data Capture systems because it provides significant improvements on the range and usefulness of barcodes in different applications. The implementation of error-correcting codes for such symbologies is crucial due to the susceptibility of high-density barcode symbols to damage. This research studies the performance of Low-Density Parity Check codes for a two-dimensional barcode symbology with high data density. A high-density barcode symbology is designed, and the characteristics of the communication channel defined by commonly used printers and scanners are modeled and observed. Additionally, the parameters of the symbology are adjusted and the performance of LDPC codes with different code rates is tested. Performance tests show that LDPC codes are effective for certain parameters of the symbology, and limitations of the chosen printing technology severely affect the robustness of the symbology. | ||
| 野口悟 | D | 藤川和利 山口英 猪俣敦夫 |
|
Title: Location aware universal service discovery for vehicular communication systems
Abstract: Vehicular communication systems (VC) have been taking on a significantly important role in Intelligent Transportation Systems (ITS), which aim at providing road users with improved traffic safety, traffic efficiency, and additional value. VC is a group of heterogenious networks which includes Vehicular Ad-hoc NETworks (VANET) composed of vehicles/roadside infrastructure, and the Internet. In order to provide ITS services to road users, ITS applications need to know communication endpoints of necessary services (e.g. IP address and port number) a priori. Service discovery (SD), which enables service consumers (SC) to dynamically locate service providers (SP) with their identifier and attributes instead of hosts' identifier, is therefore an essential mechanism. However, from the VC point of view there have been wide variety of charrenges; (i) SCs need to find SPs according to geographical position, (ii) as the situation of SCs are highly variable (vehicles' speed, nodes' density, equipped access interfaces, etc), an appropriate discovery manner is also variable, (iii) high performance requirement (specifically, latency requirement). This research have been tring to propose a solution for such issues, thus this presentation shows a list of existing works and explores solution spaces with an abstration. It would also show preriminary results of introducing location aware service discovery using IPv6 GeoNetworking for issue (i). | ||
| 清水宣義 D | 加藤博一 宮崎純 | |