設計踢擊測試機構實現SPL 足球機器人 之踢球軌跡預測及動作修正

在雙足機器人足球比賽中，觀察環境並隨之做即時動態改變的能力是獲勝的關鍵，也是讓機器人能與真人比足球賽的一重要條件。機器人需要有觀測環境、預測對手行為、預測自身行為之結果、即時動態修改自身行為等能力才能達到此目標。然而目前雙足機器人除了觀測環境外的其他能力的實現仍是一大挑戰。此篇論文提出一機構方便模擬並記錄現實踢球動作及其結果，並將結果製成查找表。此機構的好處為其模擬的結果包含摩擦力和球體轉動。我們利用此查找表來幫助機器人預測自身踢球之結果，利用預測的結果來修正自身踢球動作以提高踢球準確度，進一步實現動態踢球。此外，也可用以實現球員間的傳球功能。我們提出的方法預測自身踢球結果的誤差為正負3 度。

關鍵字

踢球測試平台；動態踢球；對球預測； NAO ；人形足球機器人

並列摘要

The ability of adapting the changeful conditions is important to win a humanoid soccer game. Additionally, it is a key point to realize the dream of humanoid robots play against human being and win. The robots need to perceive the environment, predict the opponent's behavior, predict the result of the robot's own behavior, modify the robot's own behavior according to conditions dynamically in order to score goals. However it is still a big challenge to implement these abilities on humanoid robots. In this thesis, I propose a device that is easy to test the kick motions and record the results. A lookup table to predict its own kick is created from these data. The advantage of the device is that the simulation contains the impact of the friction and the torque of the rolling ball. The robots use this lookup table to predict the ball's direction before the kick, and then use this prediction to modify the kick motion in order to perform a more accurate shoot (i.e. dynamic kick). Furthermore, this implementation can help to realize passing, which could make the games more exciting. The error of prediction in my model is ±3 degree.

並列關鍵字

dynamic kick ； ball prediction ； kicking testbed ； NAO ； humanoid soccer robot

參考文獻

[1] Bum-Joo Lee, In-Won Park, Naveen Suresh Kuppuswamy, Jeong-Ki Yoo, Yong-

Duk Kim and Jong-Hwan Kim. Hybrid architecture for kick motion of smallsized

Yuseong-gu, Daejeon, 305-701, Republic of Korea, 2006.

[2] Sebastien Lengagne, Nacim Ramdani, and Philippe Fraisse. Planning and fast replanning

of safe motions for humanoid robots : Application to a kicking motion. In

國際替代計量

設計踢擊測試機構實現SPL 足球機器人之踢球軌跡預測及動作修正

未授權

主題瀏覽