成果報告書詳細
管理番号20160000000562
タイトル*平成27年度中間年報 次世代ロボット中核技術開発 革新的ロボット要素技術分野 行動記憶レイヤ統合に基づく衝撃対応実時間行動システム中核総合化研究開発
公開日2016/7/27
報告書年度2015 - 2015
委託先名国立大学法人東京大学
プロジェクト番号P15009
部署名ロボット・AI部
和文要約
英文要約This financial year, we developed real-time behavior system dealing with impulsion based on integration with behavior memory layer using a life-sized humanoid robot JAXON.
The results of this year are following.
1) Evaluation of real-time behavior system dealing with impulsion on the real robot:
1-1) We extended sensor-feedback controller which is preliminarily developed for biped walking based on measurement of force/torque sensors and gyro/acceleration sensors to be adaptive against disturbance while supporting by the hands and feet. We evaluated the extended controller on experiments in which the robot walked through both flat ground and rough terrain.
1-2) We developed robot's hardware extension and waterproof and dust-proof clothing to improve capability of dealing with impulsion.
1-3) We developed experimental environment such as rough terrain environment and submerged environments used for evaluation of this project.
2) Environment sensing and mapping:
2-1) We applied mapping technique based on laser-based three-dimensional point cloud measurement to the robot.
We also developed measurement and storing technique of undulation information about environment and applied it to real robot's behavior to switch gait strategy.
2-2) We developed estimation technique to environment information based on contact information through force/torque sensor, gyro/acceleration sensor, and encoders.
3) Full-body behavior to deal with impulsion:
3-1) We developed real-time low-level controlling algorithm to absorb unexpected impulsion, for example in robot's landing, based on foot acceleration sensor measurement.
We evaluated it on foot testbed and one-joint testbed.
3-2) We developed motion teaching system integrating small-sized humanoid robot as operation interface to open-source dynamics simulator.
We evaluated the system through getting-up motion and roll-over motion on the simulator.
4) System integration:
We prototyped fault tolerant functionality on robot's controller PC layer such as data logging and diagnostics of network, control, electrical board, and sensors.
ダウンロード成果報告書データベース(ユーザ登録必須)から、ダウンロードしてください。

▲トップに戻る